From 7039e6111eb77374e9550b53c6f36924c370c412 Mon Sep 17 00:00:00 2001
From: Alisia Fadini <alisia.fadini15@imperial.ac.uk>
Date: Wed, 17 Jun 2026 22:33:15 -0400
Subject: [PATCH 01/22] Add scroll-fade hero post (RSS joins OMSF)

---
 _posts/2026-06-18-rss-joins-omsf.html         | 399 ++++++++++++++++++
 assets/js/scripts.js                          |  10 +
 .../rss-wordmark.png                          | Bin 0 -> 9241 bytes
 3 files changed, 409 insertions(+)
 create mode 100644 _posts/2026-06-18-rss-joins-omsf.html
 create mode 100644 assets/posts/2026-06-18-rss-joins-omsf/rss-wordmark.png

diff --git a/_posts/2026-06-18-rss-joins-omsf.html b/_posts/2026-06-18-rss-joins-omsf.html
new file mode 100644
index 0000000..739f6fd
--- /dev/null
+++ b/_posts/2026-06-18-rss-joins-omsf.html
@@ -0,0 +1,399 @@
+---
+layout: base_page
+title: "Structural biology has more to teach AI than coordinates"
+fullwidth: true
+---
+
+<!-- Extra display/body fonts for the hero figure (Inter is already site-wide) -->
+<link rel="stylesheet"
+  href="https://fonts.googleapis.com/css2?family=DM+Serif+Display:ital@0;1&family=Source+Serif+4:ital,opsz,wght@0,8..60,400;0,8..60,600;0,8..60,700;1,8..60,400&display=swap">
+
+<style>
+  /* ─────────────────────────────────────────────────────────────
+     Fancy post hero — "The Transmitter" diffraction figure.
+     Reuses the site's .hero-section (100vh), .scroll-hint and
+     .reveal; everything below is scoped to .fp- to stay local.
+     ───────────────────────────────────────────────────────────── */
+  .fp-hero {
+    background: #003049;
+    overflow: hidden;
+  }
+
+  /* The figure layer fills the hero and is the element that fades
+     out on scroll (see .hero-fade handling in scripts.js). */
+  .fp-figure {
+    position: absolute;
+    inset: 0;
+    z-index: 0;
+    background: #003049;
+    will-change: opacity;
+    transition: opacity 0.15s linear;
+  }
+
+  .fp-canvas {
+    position: absolute;
+    inset: 0;
+    width: 100%;
+    height: 100%;
+    display: block;
+    z-index: 0;
+  }
+
+  /* horizontal editorial margin that scales with the viewport */
+  .fp-hero {
+    --fp-inset: clamp(20px, 6vw, 96px);
+  }
+
+  .fp-hairline {
+    position: absolute;
+    top: clamp(20px, 5vh, 48px);
+    left: var(--fp-inset);
+    right: var(--fp-inset);
+    height: 1px;
+    background: rgba(253, 240, 214, 0.16);
+    z-index: 2;
+  }
+
+  .fp-date {
+    position: absolute;
+    top: calc(clamp(20px, 5vh, 48px) + 16px);
+    left: var(--fp-inset);
+    z-index: 2;
+    font-family: "Inter", -apple-system, BlinkMacSystemFont, sans-serif;
+    font-size: 13px;
+    letter-spacing: 0.16em;
+    text-transform: uppercase;
+    font-weight: 600;
+    color: rgba(253, 240, 214, 0.6);
+  }
+
+  .fp-textstack {
+    position: absolute;
+    inset: 0;
+    display: flex;
+    flex-direction: column;
+    justify-content: center;
+    padding: 0 var(--fp-inset);
+    z-index: 2;
+    pointer-events: none;
+  }
+
+  .fp-textstack-inner {
+    max-width: 760px;
+  }
+
+  .fp-eyebrow {
+    font-family: "Inter", -apple-system, BlinkMacSystemFont, sans-serif;
+    font-size: clamp(14px, 1.2vw, 18px);
+    letter-spacing: 0.22em;
+    text-transform: uppercase;
+    font-weight: 700;
+    color: #d14d57;
+    margin-bottom: clamp(16px, 2.5vh, 28px);
+    display: flex;
+    align-items: center;
+    gap: 14px;
+  }
+
+  .fp-rule {
+    display: inline-block;
+    width: 26px;
+    height: 2px;
+    background: #d14d57;
+  }
+
+  .fp-title {
+    font-family: "DM Serif Display", "Source Serif 4", Georgia, serif;
+    font-weight: 400;
+    font-size: clamp(2rem, 5.2vw, 68px);
+    line-height: 1.02;
+    letter-spacing: -0.018em;
+    color: #fdf0d6;
+    margin: 0 0 clamp(16px, 2.5vh, 28px);
+    text-wrap: balance;
+  }
+
+  .fp-subtitle {
+    font-family: "Source Serif 4", Georgia, "Times New Roman", serif;
+    font-size: clamp(1.05rem, 1.8vw, 23px);
+    line-height: 1.5;
+    color: rgba(253, 240, 214, 0.74);
+    margin: 0;
+    max-width: 620px;
+    text-wrap: pretty;
+  }
+
+  .fp-logo {
+    position: absolute;
+    bottom: clamp(24px, 5vh, 52px);
+    left: var(--fp-inset);
+    z-index: 3;
+    display: flex;
+    align-items: center;
+  }
+
+  .fp-logo img {
+    height: clamp(46px, 6vw, 76px);
+    width: auto;
+    display: block;
+  }
+
+  /* ── Post body below the hero ── */
+  .fp-body {
+    background: #fdf0d6;
+    padding: clamp(3rem, 9vh, 7rem) 0 clamp(4rem, 10vh, 8rem);
+  }
+  .fp-body .fp-prose {
+    max-width: 720px;
+    margin: 0 auto;
+    padding: 0 1.25rem;
+    font-family: "Source Serif 4", Georgia, "Times New Roman", serif;
+    color: #1a1a1a;
+  }
+  .fp-prose h2 {
+    font-family: "DM Serif Display", "Source Serif 4", Georgia, serif;
+    font-weight: 400;
+    font-size: clamp(1.5rem, 3.2vw, 2rem);
+    color: #003049;
+    margin: 2.5rem 0 1rem;
+    border: 0;
+  }
+  .fp-prose p {
+    font-size: 1.15rem;
+    line-height: 1.75;
+    color: #2a2a2a;
+    margin: 0 0 1.25rem;
+  }
+  .fp-prose .fp-lede {
+    font-size: 1.3rem;
+    line-height: 1.65;
+    color: #003049;
+  }
+
+  @media (max-width: 991.98px) {
+    .fp-subtitle { display: none; } /* keep the small-screen hero uncluttered */
+  }
+
+  /* ── Hide the nav links over the hero until scrolled past it ──
+     Scoped to this page via body.fp-hide-nav (added by the script below).
+     Desktop only — on mobile the navbar is solid and the links live
+     behind the hamburger, so they must stay reachable. */
+  body.fp-hide-nav .navbar .navbar-nav { transition: opacity 0.35s ease; }
+  @media (min-width: 992px) {
+    body.fp-hide-nav .navbar.navbar-hero:not(.navbar-scrolled) .navbar-nav {
+      opacity: 0;
+      pointer-events: none;
+    }
+  }
+</style>
+
+<!-- ═══════════════ HERO (scroll-fade figure) ═══════════════ -->
+<section class="hero-section fp-hero">
+
+  <div class="fp-figure hero-fade">
+    <canvas class="fp-canvas" id="fp-diffraction"></canvas>
+
+    <div class="fp-hairline"></div>
+    <div class="fp-date">June 18th 2026</div>
+
+    <div class="fp-textstack">
+      <div class="fp-textstack-inner">
+        <div class="fp-eyebrow"><span class="fp-rule"></span>RSS News</div>
+        <h1 class="fp-title">Structural biology has more to teach AI than coordinates</h1>
+        <p class="fp-subtitle">The Reciprocal Space Station joins the Open Molecular Software Foundation to bring raw signals and frontier structural biology experiments to the next era of biomolecular AI.</p>
+      </div>
+    </div>
+
+    <div class="fp-logo">
+      <img src="{{ '/assets/posts/2026-06-18-rss-joins-omsf/rss-wordmark.png' | relative_url }}"
+           alt="Reciprocal Space Station" />
+    </div>
+  </div>
+
+  <div class="scroll-hint">
+    <span>Scroll</span>
+    <i class="fa-solid fa-chevron-down fa-xs"></i>
+  </div>
+</section>
+
+<!-- ═══════════════ BODY (scaffold — drop real content here) ═══════════════ -->
+<div class="fp-body">
+  <article class="fp-prose">
+    <p class="fp-lede reveal">Placeholder lede. The Reciprocal Space Station is joining the
+      Open Molecular Software Foundation — a short, punchy opening paragraph that sets up
+      why this matters goes here.</p>
+
+    <h2 class="reveal">Why this matters</h2>
+    <p class="reveal">Placeholder body paragraph. Replace this with the real announcement
+      copy. Each <code>.reveal</code> block fades and slides up as it scrolls into view,
+      using the site's existing IntersectionObserver.</p>
+    <p class="reveal">A second placeholder paragraph so you can see consecutive reveals
+      staggering naturally as you scroll down the page.</p>
+
+    <h2 class="reveal">What comes next</h2>
+    <p class="reveal">Closing placeholder paragraph — links, call to action, or a quote can
+      live here once the real content is ready.</p>
+  </article>
+</div>
+
+<!-- ═══════════════ Responsive diffraction figure ═══════════════ -->
+<script>
+(function () {
+  // Mark this page so the navbar links stay hidden until scrolled (see CSS above)
+  document.body.classList.add('fp-hide-nav');
+
+  var canvas = document.getElementById('fp-diffraction');
+  if (!canvas) return;
+  var ctx = canvas.getContext('2d');
+
+  function rnd(i, j) {
+    var s = Math.sin(i * 127.1 + j * 311.7) * 43758.5453;
+    return s - Math.floor(s);
+  }
+
+  function draw() {
+    var host = canvas.parentElement;
+    var W = host.clientWidth;
+    var H = host.clientHeight;
+    if (!W || !H) return;
+    var DPR = Math.min(2, window.devicePixelRatio || 1);
+    canvas.width = W * DPR;
+    canvas.height = H * DPR;
+    ctx.setTransform(DPR, 0, 0, DPR, 0, 0);
+    ctx.clearRect(0, 0, W, H);
+
+    // Scale the whole pattern to "cover" the field (design ref: 1600×900)
+    var s = Math.max(W / 1600, H / 900);
+
+    // diffraction center — pushed right so the pattern crops off the edge
+    var cx = W * 0.77, cy = H * 0.5;
+    var intensity = 0.55;
+
+    // --- faint resolution rings ---
+    ctx.save();
+    for (var k = 1; k <= 9; k++) {
+      var r = (k * 86 + k * k * 1.4) * s;
+      ctx.beginPath();
+      ctx.arc(cx, cy, r, 0, Math.PI * 2);
+      ctx.strokeStyle = 'rgba(253,240,214,' + (0.115 - k * 0.007) * intensity + ')';
+      ctx.lineWidth = 1;
+      ctx.stroke();
+    }
+    ctx.restore();
+
+    // --- reciprocal lattice of Bragg spots ---
+    var ax = 52 * s, ay = 4 * s;
+    var bx = -7 * s, by = 50 * s;
+    var envSigma = 360 * s;
+    var maxR = Math.hypot(W, H) * 0.62;
+
+    ctx.save();
+    ctx.globalCompositeOperation = 'lighter';
+    for (var i = -24; i <= 24; i++) {
+      for (var j = -24; j <= 24; j++) {
+        if (i === 0 && j === 0) continue; // beam stop
+        var px = cx + i * ax + j * bx;
+        var py = cy + i * ay + j * by;
+        var rr = Math.hypot(px - cx, py - cy);
+        if (rr > maxR) continue;
+        if (px < 0) continue; // full-bleed; envelope + left gradient keep text legible
+
+        var env = Math.exp(-(rr * rr) / (2 * envSigma * envSigma));
+        var mod = 0.45 + 0.55 * Math.abs(Math.cos(i * 1.1) * Math.cos(j * 0.9) + 0.3 * Math.cos((i + j) * 0.7));
+        var noise = 0.55 + 0.6 * rnd(i, j);
+        var I = env * mod * noise;
+        if (I < 0.014) continue;
+
+        var radius = (1.3 + I * 6.5) * s;
+        var a = Math.min(1, 0.42 + I * 2.3) * intensity;
+        var warm = I > 0.32;
+        var g = ctx.createRadialGradient(px, py, 0, px, py, radius * 2.7);
+        g.addColorStop(0, warm ? 'rgba(255,242,222,' + a + ')' : 'rgba(255,250,238,' + a + ')');
+        g.addColorStop(0.4, 'rgba(253,240,214,' + a * 0.5 + ')');
+        g.addColorStop(1, 'rgba(253,240,214,0)');
+        ctx.fillStyle = g;
+        ctx.beginPath();
+        ctx.arc(px, py, radius * 2.7, 0, Math.PI * 2);
+        ctx.fill();
+
+        if (I > 0.12) {
+          ctx.fillStyle = 'rgba(255,252,245,' + Math.min(1, I * 1.5) * intensity + ')';
+          ctx.beginPath();
+          ctx.arc(px, py, Math.max(0.8, radius * 0.55), 0, Math.PI * 2);
+          ctx.fill();
+        }
+      }
+    }
+    ctx.restore();
+
+    // --- low-resolution bloom around the beam center ---
+    ctx.save();
+    ctx.globalCompositeOperation = 'lighter';
+    var bloom = ctx.createRadialGradient(cx, cy, 0, cx, cy, 150 * s);
+    bloom.addColorStop(0, 'rgba(253,240,214,' + 0.20 * intensity + ')');
+    bloom.addColorStop(0.5, 'rgba(253,240,214,' + 0.07 * intensity + ')');
+    bloom.addColorStop(1, 'rgba(253,240,214,0)');
+    ctx.fillStyle = bloom;
+    ctx.beginPath();
+    ctx.arc(cx, cy, 150 * s, 0, Math.PI * 2);
+    ctx.fill();
+    ctx.restore();
+
+    // --- direct beam / beamstop ---
+    var halo = ctx.createRadialGradient(cx, cy, 0, cx, cy, 46 * s);
+    halo.addColorStop(0, 'rgba(209,77,87,' + 0.55 * intensity + ')');
+    halo.addColorStop(0.5, 'rgba(209,77,87,' + 0.12 * intensity + ')');
+    halo.addColorStop(1, 'rgba(209,77,87,0)');
+    ctx.fillStyle = halo;
+    ctx.beginPath();
+    ctx.arc(cx, cy, 46 * s, 0, Math.PI * 2);
+    ctx.fill();
+
+    ctx.fillStyle = '#002033';
+    ctx.beginPath();
+    ctx.arc(cx, cy, 12 * s, 0, Math.PI * 2);
+    ctx.fill();
+    ctx.strokeStyle = 'rgba(253,240,214,0.35)';
+    ctx.lineWidth = 1;
+    ctx.stroke();
+
+    ctx.strokeStyle = 'rgba(0,32,51,0.85)';
+    ctx.lineWidth = 5 * s;
+    ctx.beginPath();
+    ctx.moveTo(cx, cy);
+    ctx.lineTo(cx - 150 * s, cy - 360 * s);
+    ctx.stroke();
+    ctx.strokeStyle = 'rgba(253,240,214,0.12)';
+    ctx.lineWidth = 1;
+    ctx.beginPath();
+    ctx.moveTo(cx, cy);
+    ctx.lineTo(cx - 150 * s, cy - 360 * s);
+    ctx.stroke();
+
+    // --- vignette ---
+    var vg = ctx.createRadialGradient(cx, cy, 200 * s, cx, cy, Math.hypot(W, H) * 0.72);
+    vg.addColorStop(0, 'rgba(0,32,51,0)');
+    vg.addColorStop(1, 'rgba(0,18,30,0.55)');
+    ctx.fillStyle = vg;
+    ctx.fillRect(0, 0, W, H);
+
+    // --- left-side darkening so the headline stays legible ---
+    var lw = Math.max(W * 0.55, 520);
+    var lg = ctx.createLinearGradient(0, 0, lw, 0);
+    lg.addColorStop(0, 'rgba(0,32,51,0.80)');
+    lg.addColorStop(1, 'rgba(0,32,51,0)');
+    ctx.fillStyle = lg;
+    ctx.fillRect(0, 0, lw, H);
+  }
+
+  var resizeTimer;
+  window.addEventListener('resize', function () {
+    clearTimeout(resizeTimer);
+    resizeTimer = setTimeout(draw, 120);
+  }, { passive: true });
+
+  if (document.fonts && document.fonts.ready) { document.fonts.ready.then(draw); }
+  window.addEventListener('load', draw);
+  draw();
+})();
+</script>
diff --git a/assets/js/scripts.js b/assets/js/scripts.js
index b71c22d..c51042b 100755
--- a/assets/js/scripts.js
+++ b/assets/js/scripts.js
@@ -6,6 +6,10 @@ document.addEventListener('DOMContentLoaded', function () {
   var navbar = document.querySelector('.navbar');
   var heroSection = document.querySelector('.hero-section');
   var heroOverlay = document.querySelector('.hero-overlay');
+  // .hero-fade: an element (e.g. a fancy-post hero figure) that should fade
+  // OUT as the reader scrolls past the hero. Driven by the same scroll
+  // position as the navbar/overlay logic below.
+  var heroFade = document.querySelector('.hero-fade');
   if (navbar && heroSection) {
     navbar.classList.add('navbar-hero');
     function updateOnScroll() {
@@ -24,6 +28,12 @@ document.addEventListener('DOMContentLoaded', function () {
         var opacity = 1 - (y / fadeDistance);
         heroOverlay.style.opacity = Math.max(0, Math.min(1, opacity));
       }
+
+      // Hero figure (fancy post): visible at top, faded out as you scroll past
+      if (heroFade) {
+        var figFade = heroSection.offsetHeight * 0.6;
+        heroFade.style.opacity = Math.max(0, Math.min(1, 1 - (y / figFade)));
+      }
     }
     window.addEventListener('scroll', updateOnScroll, { passive: true });
     updateOnScroll();
diff --git a/assets/posts/2026-06-18-rss-joins-omsf/rss-wordmark.png b/assets/posts/2026-06-18-rss-joins-omsf/rss-wordmark.png
new file mode 100644
index 0000000000000000000000000000000000000000..e9133fcfb20bb1863c16abec465f5614dc474139
GIT binary patch
literal 9241
zcmX9^WmsH2w;kNwT}okq;!@nTP^3_3ad&rTkilICFAl|_#bt1Z;!td`V#O&gci!**
zIF_@MtgPhOSvyIzhMFP{CM6~S0KidJlKTVzAjH7m3(-;G_Z71{P51-DMd_<M0D#;7
z--R&jTwwuEr0|gc;-Tei<>77WW(n~2_U5#8vUfK(b+P1hcC*eo6{7?I=mE-d(%L?`
zC+ofe2A2QsE~eY~zWy%5qED-l%FeQ2F5$2-ltwZ563#`RCXfH#3;C(NA&Sa)?YWXZ
z3KUZWj17%3(?llF{`FfX$7)t}_S3g7Usq*TOGekO9(T(4OO>-%PWimuHtKYZ>n0|;
zriEQn{r^bZwoSkJ9>X&PX}sfiL3UyOwI`GrFbGl!@ihbYk(-ZMahVWkAm!pbtvQOA
zH;cOPx!|AkOgnN8k!vmF1Jsb233GkwklCI)MX`J!-;|yZTN`JI1cx9M9b~Nz=lfWo
z9FU&(VcgL=bQ44Ym4RUcj{SItAT9w&ce}9ZTclw``}NYMcs#&+e2-L5Bs9cV@K_Hr
z+`{g{7KRokNrx>acRtxgXn?$?q`{PrxG`vw4CpB09)fTWqCY(E<8%j6`rw^c6bEcW
z7ocCm>I#rHP%R>yEdE2^_ZjU8*JmGgEP%YO5sV?6$O?6Y7KYU=5f~i43-}kY_|GDa
zXX^+lf%(7>5rX=JA;dC3eZvG)&50dQfCZD6DS=2Z&Es56kzAO1`AG3S5CRijBBr}h
zz5xD8-nG*SA^4b-&{S)Xo5xYvW5f)OF`O9!o-iTWWe_?>a#w-C^Ph<2L>}Py5lk@8
zQ0o=Hi_m#r(rp66-H>E?xa$!{txnLu;<Sr_!nwh4#b2vtxY4ZwOd^z}_5CplC$ms5
zSusf~7#4k8=tK_*v+MX918>!uKe7<cd1R+9UO=RY#q1Bn-^4Hj*rtO%(<x(O&b|&X
zj;D<G3G;57`)nFX1OqcVuA$}`gcf3tjya`~2W6FVs*e7;l*%)tjhhG?y4-pb@B(x}
zIj1r)flTiMGVzHsw>Xg(i9FmnCRte7iQzr{)r;wX^S5=HV?I(4mIuCAW&@rJdHtE!
zZwWABnMjL_X&g_wHD@o9ujgBbE4@&EQr{;3|MK~g0*u*A{S%Q2*gUNN*J)$gvfQN#
z`WV0c$a>q!Q5x9c^j4}qBirkc*-OY1W7PjI!Q+*&)mpQn1@H=FM)?1HUgiXkUY{(I
z0sjMZc*<tBsmbz4#s0s~m3&4>MkoGysS>?WjKa{ntm^*6v-n{N;3H@Puuf4(T7q!F
zQkefukI2r*l;{x62-oP&p5Wr*P~L}P#`?T#VPMym=@$Fqj_voVu~)XUHvg$)eD=SN
z!iK|d{-74D15PO<84fmA6Sw*XHR<R!)g|GX;M9}OWlzd}!swY<-n`rv_C6=r8RFlN
zS<nbB0GKLeM-ac`EK&OzNK2F<p`%{?&3GwAEK+krCfP<s>J%Wif&Rc8HQh%Q$BM5C
z?#>Fu{FW2ir#Tt%;;nM-c7@H!MvwUd%W0!#n&U!L0lfvr5`3K)$bKt!v;0ucSBD<|
zsQzX{6s-hjW(C+3BDl@}E$0;yUh^W!uHUoKIoitna?^q#tun6a*0AC;A>%2?og?=>
znz{kj3DBphuE8yKdcM_)zgdYyj764eCuU7+)`&vq(>ID_8AcqY|A9)kE715o8_&1U
zi~h^`lVct6GEUBeav7bBg#^+|K!!J`SB@D)^ng<Hl>BqQG%ic6JA(7L@_^h}u5jvd
z{ULt%8%<m!XRd*Sy(nBskH^)xzg^IKY)P(pNV=7e1FtYjd_lxW+qXfTdW61H!YbS<
zjeB9ArMY#Zlpy7OJ-t6kv3c<aVkRt0huVdbDU-yWq?Ap~&3b;%<`++V8or`8?<E8;
zgq{xZ>i6>pf4EV*DDvPC>jL%kjKa<gKh&ogRX(5x{$=*(34j@C*Osze>1(qW`<qsW
z5{)TG<>Pd&c#3?keF7I%I&yWctw1AJ8lTE2s#4WM$0dy&*iVp~#z+G=fz=ul5T1~S
zh(rW31T8EaiBP*USL``PZAoSC{lHqH6;1+%88>PlV-!cXyeT2gZkd_IMJeOcROJ0A
z-MUDVA+t*h$gAkGW#*?JUZgI8g`)eflU=@>?>}vx1J5l#7&Ad*<XBG<M+%h7f>pHI
zS3G&&*xC7tN6&}Rx9BHF2_=(9U*BKv(V|Htl0^3zx@lI%AVZfx>zM1lf59>XZ?aw(
zk0g<`PnCzU_km)h$w~!VBFQyT12#_2@A7ZGVSuGW2;*C;N&4h=ye+oB`JY^I1K(4>
zZfr>&pMX1p(Ds9fH$2Iy&%K29_`m!(;51o4=TJim!UW@B(CIH}@qUdQ2#$06#2~GT
zLU39l=p}M;!X-5MCTopoTM}vGlt1yJ#=}rOdeI#J89|3Wre%7=WuL3C(De8g4HZV#
z()V+zw`tm`qhM^*!+Q-O7r_AauZZfgG3C=cuGO9*8+&`u9bJu*HpbqU2e-r|)%N3U
z+YgiAJDF-F?MRD<*r<(YlIZP)rT+BIjb<ea5z`H2mwS?avS<!Lg6h;f#9Fges^v_?
z;oY3Y18`AIO-A<`7sk+e&QUVxZd0kbTkvV5Qucgz8ah=tELi%M@Xd@L$GPxQrFQ&y
z51>XipIZY=6@)%ai+AG7;oe2siZ6so{@b4(iboNcwcr?M0F*PkzL!Mmqp)`D6QsEg
zN;wRZiG`@<fkx;+tc8R8dpzDwIMMSpVf~Pgthw%_)c)5QOUFz7D6d$?uU6c$vZkk~
zH%PP;mm3MtGYr~Yn$%=0dB%SNzkI<~){Lf@E@T`W&OrZ?KiA3hl3{iQ`Igl97hhUl
zm_yOpR9UHgk)O`M`bJ#FxWjqqS7)s9$LfS0S4IN6T{^I(I9Ki8=kW3S4;{86&um>(
zh_Km_W1(-q!pJn|tk4EFu~j7U{CMcRB0f&&H=YAKHb$Wc;Dv?s^zjVKs;jdWUlVx9
z=S^okk~$*=O^}~{cJ7ciTf`RY!tl>G6<fh@yZtratUIdj<g|)+hs0gE>zH%?R__&W
z2CJo~kRnwZ%OH?s9y=(qeMe2`QEa)j!GNF+^$$W<Se-g@cSBcRGY9dkv-!8-a`6k)
zp#C@tq7RGP8%^F)?%o>tXnY@+Q!0xIp5u=BzYiJt_a-`_=R;9j56fLHtV2!f_dL;D
zxX;&M$B9Upan1dUV8$#7bf&XHtunolpn3I|;8NQU;lD$uLWWX)Sg+aHeGGj<KDGtM
zAI@#DNQUQmqi*6wbiHh5hH0v8!Bxre;b`xYF`@!Bq`nmH<utNW!=t&yBQ1{7igJ`o
zI1$AC9;vcp-kT30i@gf|9joUr`xr*|$_*h9%nVbJ#FFe<k!PlWNqvijVL=7BpSsF9
z;JokA(?7ZXD?=Q;wXt0(hOOt!Hi5X(-qGyn1oo5AKenkAv#CM=OM!TVAAfD)kp)o)
zeV)gL4f=(`#<n#>{alF7^L3s1uVc2a>v{P|oH4tJy*ydBw_Iri`zl3{<zEAqY_-AJ
zly@86TR$KBhm3@-s9X0FYnu@7xO&B7dkCJ^H>Rp+8*wq^^;Z*+*%>JKl+>8Are00Q
z%jOT4MF<xdfDfC)F+mq*Ja;$mx1dNS>ni7kiBQ>K_q+uSdyqHMEr1BrK|b40OdMnM
zYPjqHwbIvR;JsO4P90@MH)b`LU{?Kjs1?%E3U5llmV0kp{j?-rc|Pi^(Sb<umIEMx
znVdub468h)sBhEo@r$aOP?fhYV-0*-exF#N!YvMp79qf!i8UzKBvBJ#xNET^$$fe;
z3GEqMK+BclrknSG=}HA^0EyX&t$d0yu@j=G49(5O$50q7Ec%jEX;io;!&nu*qeysD
zh&`cGi{j~2yX4J$*+tK2=>{y*-9;$!w(%voWM@=bxI?le0y^{uk0rv1yF)W7xyJ@;
zwJI<xNn0QE?qG_rK{7+@)VedmwcUJ5Mf1rU^mvN&{Q~J<#=_Zr{Yr!3^pK#STKG`x
zs|!9B2{RK%7TeugX)Lqy-*w5(RtQ>VOSfxqz4<KnOk`MY!WJ9(Vb+tdY<q#>=Nt8t
zl+;w&iNUngC{Qj;1V`(u*e`KI7E?y(+>9D+vy<{qxEn$kb^gbv^}Jp2_u!6pFcumg
zwWg?606m**G?PE`y8SL#Id37jpju)z$Ar2Z4ao!;#qwsFNMy9mJ!x+Y^`Z^8y}s0x
zH|Gl5i!{WL(=-wI0KP=k<NwRp!Kog+H8$HGz=6vGza<QX(1&zh#7hjAv<bua*rdDQ
z=zYQZLMskFZ$I0dO<64+*MYrO?|Yn_Z922P?GYR_VB6GGCMXltm;>zv<p~jF;bzV6
zCBaz0<YPArM$PxLBCWTn^6tt#Mr$-et_6kAB~8}4=c)*}JZnGq`&@o)=M7fq&-t(r
zHzk`#67#s?{&1ngG(?{hm~lEa&Ct*SO7mE$dOC2C@7;^GWZ(&;S(#|5R!WmP?~JS3
zsRZisc+ReVS#+%<i5}Z#P<6*&iVWfVd&#42c#`;OzYF2OB^AmImWkk?HT3~)Wa5=H
z@e!(X%lA!a6Z*ubE*I_3=CviEU_Imwyf40V;oDj&7_p)W;!CK%xgk5>BM<fc+LuAO
z<G<IOEi`<wn`)`g<z{d>KiGi0<<~CKbH>z6LG&Ihb3I)+T+J*_^A#qF%XrqbP^E3h
z%b79xSwfNdD&WAiDlcb7jcdl(+5N+3Aw05TTWwk_ajT!vhXT;Dq~cSdFVu+7jp`#8
z2>Hcldj-N#`p@&djX&8W%?67E3Wo)JUESpims@(xJiR1vE^~hE;Bw&Rr#gR=Y~U<!
zTw%G6+BT%~Q%!yUUYFQ4Q1}|Mp!Bb<aQHex^{HHq86f~FJ>e?TCy_rr+r>}hrm6*S
ze#}Xe?pglT@<}65=n2AmkzOFr45pz$?R#~mm1K4bPEC1;#0iT>eGl$Si;yf7y@urD
z3TN4j;JqH|HLh6q#|!5xErS`KHjTCL*@-o8?uw7mNhvTAr!%V1(_OH@G3(dws*<pJ
zmM`Kk;s48ykhV6?DE#CZC{N#O_*%vgY0)x}1&QWsnt~a}LRrcBy3}E@lDnUmpT|u%
ziFo#yg2Il8CzihyMm1*fmE+%32sl>}*fS&=X+;u!NkC&rRM1{wnD;3&(E%1sR25M8
zsalSEtQRq;uXAYG+Pd<Q&9uV626v^bYd&Qfa2r)OXDsnD$z6*gqh)(B^tR;?hGA;O
zKF+5H9l5{M#0XwVMXFu0vWbU>8Ma&U{RNC0bN#Z3!a?DIV%|(E4-ZM^*@$p3-~asK
zx48o|k{Q2@O|xyOU)1C(n6K6r4%5)l;)xd>vB)IaURr?}W8{$I1LMyVH~n;eNH*<R
z_QY0hA4l(Vy5oCOyEFhZx-&B}-iOmI?e=5J6TkgLhNLJQ)frI5N)}pPVp#pon>P9y
zje4NQ?d9^sS<8(l;Qnp&04J_l4o{;Dfk5;%3y3{;Aj7oWz@-V0?m{zpn3^|Gt3x;~
zf+>bn*_Js_P2seibJ8F~KvFq>%B1DZ7(k>u{BbYYadW2tmlvTzwk_+=wI~1SnVInB
z5npK#tCNlzG6anMu0;5`yBAIQBT9pe=BS();bBXXO;<+lHHdnLpHA|&#c?yOC$eB9
z0|vAvJTp`-`ghFVa<b2zd4RQ8<@Nav!6?j)Z*;zxNngFi+KMgGhLPljzej@O+M-=1
zQJ+;#>?T_vJsU@0zzpZ^0;9q>(QoeD&Hv4YYQQ?)77lc(c8+F#Sn7;lr}|+1XkAVp
zaa8!Ik$jOYJh<qcf~9Z~yQb05xT91F-WNq@!-VwMe!AaOD5KYzAu0ytJV3N|<XvMa
zmcAWWP2VkAQi#B=9qP#75|75}@OPBj+VCM1c4G4dxVA1}X2?XYBk-_J%ft6V^rC!f
zqlNHV)Y)Yh!OkrB(6Aqsd7tT^5!JZUXp6YMU8u`ZM_1_6VYi55S5dVJmMDyd^_4je
z&C|-wA8Qn0hi7>M2@&zoKU^v~pnu<vHGFXs_fd7v0%*5v114oYX5`<L7j>U1+*Pg0
zOKZ((SBL`CH!(X2Y!JNgB#4c&u$*nw?G|pok2x#8L~R!3Qkd~-Za4fRi-7Ixwr_ck
zpC#CwXSb>|J|oMpoa4djuDhR<fGW@js<Zy8U<^p-y5n#;h6CStt%Ii?sJB2KgrHO+
zn4;<(@=p4nbmv-_ho`)qrDR9Za&@UCZrd_zR5uX>XTYFbTQk76tla#Lv&`aQ9V))X
z%zx^DNJ`I3ck(K{;&88!^H!?`l>4Rm1_#;%#AEbBBRUS6G*1MJ)2JB@c^4{8WQSnq
zV&)?LJr>D}dHb%6ozHq;dVi|M^yFayG<iyii$aUqmesl;19yPR`s!tWLe_L<1t)Ec
zQ11nfHdUt^C2?!{v^0nIE6P3m=u#@_=#j*snGK9QBF2i6w(q|D!t+y9T7i5k-kJQS
z%Lm<I(lX{aL@$i?^bX@-BQ8SyrA2U$6g^K!Si_?8+aj7IU3KNZASJ6iAE8}J0&%`o
zjTQ0X+&-xOdD3i?J($gk9EYB4%b#7TtY>1QDV9!(z$5Ib<ZfOSL64vdEN90vI)71{
zy5jIe93yvZI|BU#!>*Q!%X7gR&wTRkShdXLl=;_-&&%6y58y0pmXzi7^q8r+cO&VI
z;@BDLrf*4Aoo2~eoAcydFa}btdcDnh!6}mLRdCzoD%VZ))!*u{oEV6k^hh#Jb?B<8
zsEW-`co&7E;c})C)sm_tO7r>4I+!t81mfewdM;V36<u+E?ZHWYwdjB9#iCRa@%))9
zPtozUN4fbY=~_68F{`OenTr-dEj)U|aCH31Nm#KVGm|TUo7~ZmECjyn?clEyKkW$y
zTo>`~nO`oUBL%<73i6ls=hA<m@5_5THA;NS|KN5Nw%^(JIS*VU?8z|ZDAjeWU8EAE
z>UCJu={;_BD$~}2jgb{P!VjZ`lT7lX12-TBMuDqEfjd+c2>hGnRaK|KU(PAJC#t;@
ztnah9y(gi2zOfxhdB9~TQl{h?wV}%E`C-E!74IhgSt1=nmWuZrP*c+ZCaG^%D;u4!
ztS~qgi9eXT`qJ$1xwpd_&hUJKtmTE2Q%T!D!}~K47uNccU5YF-{@J?Q@CBUD;{P#(
zy{j<PTdlHJ+LCh8*>+%cvwTv8d_|l!rWtm0!e45=_fuDJ741^G8teP)L|;``dKez(
zV{YJrCh$|`I>9}K;HYVGN#w_$`)MVWk7snCkXxkEJfl8hd$#<wF?fj4+;fP(k^t78
zy56yhmG!mr3<^`j#DA7yV7yxc{M{8vkMhrlyymCW&LvIW!0r_RB;s2ucg40xRNeWv
zPH=6a(bfp3Z95?2PvGX4DZmUD@yE(54dFgdN`e-GT_q9G%=FGZ%kh`!vBwaD(N)1g
zp+t*^U!Au{qXB&p6FRv<N1Fq*?kpMXQa`Zd;qtVPq;@W<^=AdGa`Vnsqx`{XE78_7
zC}+@T!Ug$`(|!NX`@DF&gqMw5gdR2(oidF{)!*mQZUVF?GSW|1QU`bi+BkcXN`L9w
zMbFs%t7|8?K>bonY2q0aRZRAa1?jg&>^j%8>%w>oPaSot_Z30eL(PrV{1odl$sc}#
zYQF1)m)iXO=<!pMQg6WF4;Q{Dz7OLMh?Pg*J{F}}<#!Knqws^_`cWiyc_g0Kt34Z`
zb}-|<VZ<b&KDvBcLi=*gfi2!^c%ET9@kwykKX=`@C3*om4xh8yDfWw9jp{pH)R{S3
zyVa`sNrfhjAI+zzexb+wj5Y!;{Oegww$8yTUrE>3r+=mUp22v2Y&F$J=jzW}xQ~mR
zf<wNdFJ*GGYe9rIplzj2{Mq0l<7vd)i6U~%K&SxreEF!=saxIzdt+Y9&b>RRz<iyI
z^JJ=Q%~k#<T!I#}yOT01BC4pQX0XQPY90}2%&-&pzAt!khQ_PsXxvLss;5i_q>;q3
zohe*4T2_r%sY$yzqPLi!^yBP3rTmy%xbSZmwJvsR{3v`jk)Zme;I-4JILfH-;3ODL
zBSZ9YLAC(-nNrxctYe3|JyYyZr1<3MXzW+IFh%DIQ~>H;t3_$Op{g=Ik@BPDXl~sV
zYduR)TMh9K64G1S)wp=>DJ-y(Yv#zpG0HDGM6|gs2fD^QvZhY+(&LiL48!&`hd;%t
zc|Sz^G{bZ$PEX6@SoF4gYQCAQeGqgPYP-wtMQ6Qd@me5{KvHJ%V*PAh6rGmwy@BV-
zwvc5m{Wb`Vj(Wdx;T6G1>!F0vTA&`V%b<N$TJBEpec>Jr2w&yMmU%|P_hmVd{%>CX
zE|Zuv*;)Bl^Hb$$#jQesIQ||WyamzIGEsB`Xg!n1ItmE0A&HSY!k2q7h=^m0SLdd*
zIWMS|8QGo#|B^Rovf$nE|DgfbO7nM<K;?Pop;V@^T4eVi#~16%VxO0|qB02Az*?I?
zXKQyd`6O_%f{DNpIS;q-O>aXkv7Oq2sl>{}R;dk8uc;=6xVgmq9+-rZfxsS8PB6N4
zOeu#3$^=P>IrrE5??1)UcL9_f-WUaQ2?*HRI0rN2UcYgWf3yp&`rd>4*cPSeew6?v
zc58_K*86@D(z3VjHqJwFu><=k&mS?yKTy}3DNeA0<zf77*L)+Ee%H(y#R)^;C=}&c
zY6;e}B_Oz{rQboMp*Xd=m40j9*dO(nmQqK1pe<*2ZVR3Bo6iA=pjdGI%7VT#aRc7R
zO(H}MXmI%D|Fj{&^l3Uq1{wWY)jY#>^9-&WNtTP~b!y$OxX#(;v;N4k_^NjPCxjMb
zAotH3KY%G&Ov9IkA~N*b4=l!vJe2x3gFC<hQ560Yq2f^{J!_<UF54%z8CGIO9>U5-
zwbBk6iG$nG1%s)}x38X%$V%qkv)M`LB-3xgq({m=e;O+fQ2+7_X@I*i_B(~$asJXo
z>FZVK#;MyS-Zx}0qeyW+yb6@ER*YKv^IeU`y#24q+?I3e8y~&@2+2vtbu8>!@mMFU
z;{h|i&#t?&@Z_O9*jxK5+4`9lX#vejsO63D8e~IlG2<|#b@$_;7p)Wul-A~)hxAUK
zM}GsNPc4VT)=q(8DY@^9wSY;pV%TY1>2sr;91pXTkWOT+nt+*)=eapOKkZ7hi>DsB
zoyr7l+2r34uo3H}yu!rR6}=Vr)|dHqWWDz@rTk<PkIlhF^P$Q@lBe{S+P{B$y=zWJ
zC&5J1CH}sNE?&lyQ~5{3vtHq6>)j`v=f*6_PoY%fzN^c|*B5abh3L=9Yo6Ed9<@ez
z;#c%INrN_<D<%UpIKSg&B*l{ip2mGDv_TihK9DW-(HMjjA>)yHMN$i__58dTxT3|D
zEc>sNPlf2-Bl(R?^FBjya$l+420r+;x3*s|fdk(aB7-GAfEjn$G4*{XwWEFolP%;I
z12c!|3{{r3YRB)`h_`AJPsV?1>`?F#rYDt0UpO!?xoLi;UvrP;lCMLhER~s|vg1!q
z;e2bO+B<-EPt}bp8C}Yn?vhwOGGvw$1+V))Q@0#UH4P!KQQ|A<CY;8UU*Q*AMEX}a
zMBH<*V+%HuYgcT&h@%Lp_iw$3c}2BTz)y?}Ho8#!@`oNzlZ}|+^)hBW%Pt-xY5=qZ
zrb>lpq|!CpGnPJID7j-F0FuPiFOk@kR@&PAkE8m0sn$0y+KvOpD@u3ZBINxkxW%w+
z0$G2R=-XY$ki*A+NL$0$&(4kkO3cR~-@@UV<l$JyKFq<2Yx5|%!`I|WX7~byoFuuQ
z0X(*#6(~oXlg0h5!rauptWe}mNF2B(J<l|nao#gSft0kvCv~@8&|QlS50UNd?E%km
zVI%6;5*?ap2gI12=jHmS8uz$-T1|Naf;+lJdoD@pBGD&H;o>dYo<szG?n_AVTf>Vf
z)Z+)lF&F}`%zqU(k+h;l3zSN5z3^A#6ekaF88%L&CZw;^!-xp7KSB7c38Uwy@zuNt
zjrWp2@e}3ATr8U(hYVB8Z=xhi&aC;Nj9M(u<>g#3rrR-PwIl3V^#%EP^X;9fZR=1G
zFl0%j;pDtvbWVSz%}?9)8B&>1vI*}gW=Le+WQ;hnrhidRIn>uYmHj$m)M3^78bqcm
z!hA$Yb$nrG&9L!;eum(S=TyrW(I}%?A!kMw)kB)t#$T$2qDO6=Ci|3Ppea|SQ6Yis
z*1hxY2sDSUCJQRpxlLB>L2sn>z&R&8Dv6!Yjt~srVlmZ1W=OE3P9|??r<7ybeNrI1
zQZ(X%!^7MQ#c!%~=(G5SU`g1AtMrlkTq<Nt=zHSq%Kax+${3wO$d_AxK?l9Ws`1OM
z1%wXN9-1(=oy-C4?8zRo9D7;t0(uf{1_xD*|E*lOcSMGow)St+wIYkEL-~DL6zq-i
zqz=P!Q`OI(f_S#Ad()>y`&5RAong(NeiGNnAkh0&9TUe@etPrjC{SvI^_S9!tqj=L
zGh(H3qa=NeIXIMg#{A@45Lf9HdvIz)cW*At)2XdG6X3*V=?<Th)g{^plB*2h<$KQ5
zCEO)P(fi3njY77?rb~WPsjQe>W1>jap0V{Asr`a{e=F|6MRh;4Kd~qa>gpdKCYJxV
z8aADmuF8)kuRg7(UF)3}!Si{Wz2c9;)BF_|HV>IruIDbZf!2{i@mC<=VaFID-xY2`
z&=Ourq<A{5_Tw$+Has(dHO5VtowSD4vuTEdn4PRA>e6<}mYm2`Vr&H-sz`2=8lCbM
z1>}<(XK*`y`g*#l@JRNp{E8VS@t>7)AKt-KoS3B|+d`;WthapsO>7WBks(l<q`)1N
zB39-!A=H@CogJ&V&bS{KUopAC?R5CLnDm)ZzC!~551yse|3ghMi<dhwHprFGDz9}+
z@J%r?GljM9&HNx2n>FZw3=`zA?^t1QD1u{NGzR?&KMR1A{oh#tX)dOm`{5|M@=vK3
zPiK*kB-S;o-CeogtMAkKlst(sLAFGaC{GKXltbG5d&3r@bu^wixds@#Z%X1+w*<HR
zS?W8B|EetMs26^z|3-(JfJoY|HVDD|CtCJV#C@J4rIHAQ2To034NV+FJNWvr&mH<%
zy$~>BwbTnW1&$k(o2i0*A<;is;zih))9n!27zT1xdv_7rMV&C%^*l-7uy4NF31I^t
z_aHn37tW|hcyyk6k<HCCqoan<wDVbB6wR(U2gT8XZrYFk;O44-T<vxzTlLLXAs<{N
zv8j2F=Gxu&7k_0wk|besacBYPcE#F7dx*VYNAHLMG&HZK-(Kjo6^(w+K<t>;My1XO
zq%;32tt!QN9y=rAVsdb4MSZO3OmrB&H9)T0kU^+7o1uJIq>C{jLeIXi0NjbKOznJa
zDDobtkqeg__XM(lGqYKtA7F4foy+lCXB~_f`j5Sx*Shf12!^tBNUmk{z?AQezA~{_
z@$klpd3z?UopkpN1GDwztqSVVE!DraOnmL&!sY6Y9W}R0y|m2^IGw!bAyg~cqLG#=
zif2xa_Em3_Ur_qKZk1SJo?{;<j#&LkHy2zEh%bP{>okA6p7lmE5Uqa1Q)DP@BoH?F
zHs}7lV$KclSKJ@%!L0{LzoAyf5xE-?Nx3mk5-xzPcHH~wPNyF6vsAPmwB@(}M*AXf
z%`Z!$HWa5GNyTZr6KxYSGKA;-ya9aUIq=qrz{>9$>gMKN!jCPC90N!(P3m87m|@I>
zhf7<oe_&~2snph16E=&CYcX7#Z*y$v=4csZ7@rBwYmu2V-ixP}598OC3VQLuH~Ko%
zteD`+GlF)Z{4M0)&pxJ9tzX&RutQT|B!YjN+9>%qBKXEoLC8i=f=J5Yf}WmL%2lnb
zkM&9;-~EBCFFqvE=W3MG@JOVvp<ae3d$;?M@g<~UN+An8;$ZUq_=oCee^g7JKC`sJ
z`Ex+l#Ex&vEPRFM3<{1ih%uCJO_4oRu;AI|V&3N{Df|<nS&&qxvD0#F`!3Y3$8)e?
zrTsmRqqL>2n3w$Xx|A2f4S-UrECf@UOy>k9c|1i~u!tZWN|n}yJpH~i{F0HeO8OP6
z5%bAfBcGXUy1ppTQ<H#->Nr|mi0Yb9&WrU$vKM=dfLdK+k?{_SY9fA)T~B$3H^K+3
p(s*SRtT7%-Rd@2paK(5I4zs_aGy+Av!jJU;l;zdrYGq79{|{VJ;Tr$|

literal 0
HcmV?d00001


From d42e71785c27bb8a2d0c6e45f0551dc426c5a1c1 Mon Sep 17 00:00:00 2001
From: Alisia Fadini <alisia.fadini15@imperial.ac.uk>
Date: Wed, 17 Jun 2026 23:19:03 -0400
Subject: [PATCH 02/22] Blog body: content-body styling, RSS-blue section
 titles, scrollspy TOC

---
 _posts/2026-06-18-rss-joins-omsf.html | 108 ++++++++++++++------------
 1 file changed, 58 insertions(+), 50 deletions(-)

diff --git a/_posts/2026-06-18-rss-joins-omsf.html b/_posts/2026-06-18-rss-joins-omsf.html
index 739f6fd..89f4e23 100644
--- a/_posts/2026-06-18-rss-joins-omsf.html
+++ b/_posts/2026-06-18-rss-joins-omsf.html
@@ -1,6 +1,6 @@
 ---
 layout: base_page
-title: "Structural biology has more to teach AI than coordinates"
+title: "Structural Biology Has More to Teach AI Than Atomic Coordinates"
 fullwidth: true
 ---
 
@@ -138,37 +138,17 @@
     display: block;
   }
 
-  /* ── Post body below the hero ── */
-  .fp-body {
-    background: #fdf0d6;
-    padding: clamp(3rem, 9vh, 7rem) 0 clamp(4rem, 10vh, 8rem);
-  }
-  .fp-body .fp-prose {
-    max-width: 720px;
-    margin: 0 auto;
-    padding: 0 1.25rem;
-    font-family: "Source Serif 4", Georgia, "Times New Roman", serif;
-    color: #1a1a1a;
-  }
-  .fp-prose h2 {
-    font-family: "DM Serif Display", "Source Serif 4", Georgia, serif;
-    font-weight: 400;
-    font-size: clamp(1.5rem, 3.2vw, 2rem);
-    color: #003049;
-    margin: 2.5rem 0 1rem;
-    border: 0;
-  }
-  .fp-prose p {
-    font-size: 1.15rem;
-    line-height: 1.75;
-    color: #2a2a2a;
-    margin: 0 0 1.25rem;
-  }
-  .fp-prose .fp-lede {
-    font-size: 1.3rem;
-    line-height: 1.65;
-    color: #003049;
+  /* ── Transition from the dark hero into the standard white article ── */
+  .fp-hero-to-body {
+    height: 120px;
+    background: linear-gradient(to bottom, #003049 0%, #ffffff 100%);
   }
+  /* Everything below the hero is wrapped in #content-body, so it inherits
+     the site's blog typography (IBM Plex Sans, blue h3s, styled links/lists)
+     and scripts.js builds the scrollspy TOC "section slider" in the sidebar. */
+  .fp-after { background: #ffffff; }
+  /* RSS-blue section titles for this announcement (site default h2 is near-black) */
+  body.fp-hide-nav #content-body > h2 { color: var(--rss-blue); }
 
   @media (max-width: 991.98px) {
     .fp-subtitle { display: none; } /* keep the small-screen hero uncluttered */
@@ -199,7 +179,7 @@
     <div class="fp-textstack">
       <div class="fp-textstack-inner">
         <div class="fp-eyebrow"><span class="fp-rule"></span>RSS News</div>
-        <h1 class="fp-title">Structural biology has more to teach AI than coordinates</h1>
+        <h1 class="fp-title">Structural Biology Has More to Teach AI Than Atomic Coordinates</h1>
         <p class="fp-subtitle">The Reciprocal Space Station joins the Open Molecular Software Foundation to bring raw signals and frontier structural biology experiments to the next era of biomolecular AI.</p>
       </div>
     </div>
@@ -216,24 +196,52 @@ <h1 class="fp-title">Structural biology has more to teach AI than coordinates</h
   </div>
 </section>
 
-<!-- ═══════════════ BODY (scaffold — drop real content here) ═══════════════ -->
-<div class="fp-body">
-  <article class="fp-prose">
-    <p class="fp-lede reveal">Placeholder lede. The Reciprocal Space Station is joining the
-      Open Molecular Software Foundation — a short, punchy opening paragraph that sets up
-      why this matters goes here.</p>
-
-    <h2 class="reveal">Why this matters</h2>
-    <p class="reveal">Placeholder body paragraph. Replace this with the real announcement
-      copy. Each <code>.reveal</code> block fades and slides up as it scrolls into view,
-      using the site's existing IntersectionObserver.</p>
-    <p class="reveal">A second placeholder paragraph so you can see consecutive reveals
-      staggering naturally as you scroll down the page.</p>
-
-    <h2 class="reveal">What comes next</h2>
-    <p class="reveal">Closing placeholder paragraph — links, call to action, or a quote can
-      live here once the real content is ready.</p>
-  </article>
+<!-- Soft transition from the dark hero into the white article -->
+<div class="fp-hero-to-body" aria-hidden="true"></div>
+
+<!-- ═══════════════ BODY (scaffold — drop real content here) ═══════════════
+     Mirrors the content_page layout: #content-body gets the site's blog
+     typography, and #page-toc gives the scrollspy "section slider" sidebar
+     (both wired up automatically by scripts.js). -->
+<div class="fp-after">
+  <div class="container px-4 px-lg-5">
+    <div class="row gx-4 gx-lg-5 mt-4 mb-5">
+
+      <!-- Scrollspy TOC sidebar (large screens only) -->
+      <div class="col-lg-3 d-none d-lg-block" id="toc-col">
+        <div class="toc-sidebar sticky-top" style="top: 80px;">
+          <nav id="page-toc" class="toc-nav" aria-label="On this page"></nav>
+        </div>
+      </div>
+
+      <!-- Main article -->
+      <div class="col-12 col-md-10 offset-md-1 col-lg-9 offset-lg-0 py-4" id="content-body">
+
+        <p class="lead reveal">Placeholder lede. The Reciprocal Space Station is joining the
+          Open Molecular Software Foundation — a short, punchy opening paragraph that sets up
+          why this matters goes here.</p>
+
+        <h2 class="reveal">Why this matters</h2>
+        <p class="reveal">Placeholder body paragraph. Replace this with the real announcement
+          copy. Each <code>.reveal</code> block fades and slides up as it scrolls into view,
+          using the site's existing IntersectionObserver.</p>
+        <p class="reveal">A second placeholder paragraph so you can see consecutive reveals
+          staggering naturally as you scroll down the page.</p>
+
+        <h3 class="reveal">A subsection</h3>
+        <p class="reveal">Subsection placeholder — note that <code>h3</code> subheadings render
+          in RSS blue, exactly like the rest of the site's posts.</p>
+
+        <h2 class="reveal">What comes next</h2>
+        <p class="reveal">Closing placeholder paragraph — links, call to action, or a quote can
+          live here once the real content is ready.</p>
+
+        <p class="reveal mt-5"><a href="{{ '/blog.html' | relative_url }}">⬅️ Back to blog posts</a></p>
+
+      </div>
+
+    </div>
+  </div>
 </div>
 
 <!-- ═══════════════ Responsive diffraction figure ═══════════════ -->

From bc7283810b2d46e3379763ab2c18fae8e66a73bc Mon Sep 17 00:00:00 2001
From: Alisia Fadini <alisia.fadini15@imperial.ac.uk>
Date: Wed, 17 Jun 2026 23:28:39 -0400
Subject: [PATCH 03/22] Add blog subtitle for OMSF announcement post

---
 _posts/2026-06-18-rss-joins-omsf.html | 1 +
 1 file changed, 1 insertion(+)

diff --git a/_posts/2026-06-18-rss-joins-omsf.html b/_posts/2026-06-18-rss-joins-omsf.html
index 89f4e23..59ab378 100644
--- a/_posts/2026-06-18-rss-joins-omsf.html
+++ b/_posts/2026-06-18-rss-joins-omsf.html
@@ -1,6 +1,7 @@
 ---
 layout: base_page
 title: "Structural Biology Has More to Teach AI Than Atomic Coordinates"
+subtitle: "RSS joins the Open Molecular Software Foundation"
 fullwidth: true
 ---
 

From 66587f7185627e4641f8c06cb94ea8896a76b5fc Mon Sep 17 00:00:00 2001
From: Alisia Fadini <alisia.fadini15@imperial.ac.uk>
Date: Wed, 17 Jun 2026 23:52:38 -0400
Subject: [PATCH 04/22] Move OMSF announcement to June 25 2026

---
 ...ins-omsf.html => 2026-06-25-rss-joins-omsf.html} |   4 ++--
 .../rss-wordmark.png                                | Bin
 2 files changed, 2 insertions(+), 2 deletions(-)
 rename _posts/{2026-06-18-rss-joins-omsf.html => 2026-06-25-rss-joins-omsf.html} (99%)
 rename assets/posts/{2026-06-18-rss-joins-omsf => 2026-06-25-rss-joins-omsf}/rss-wordmark.png (100%)

diff --git a/_posts/2026-06-18-rss-joins-omsf.html b/_posts/2026-06-25-rss-joins-omsf.html
similarity index 99%
rename from _posts/2026-06-18-rss-joins-omsf.html
rename to _posts/2026-06-25-rss-joins-omsf.html
index 59ab378..df00796 100644
--- a/_posts/2026-06-18-rss-joins-omsf.html
+++ b/_posts/2026-06-25-rss-joins-omsf.html
@@ -175,7 +175,7 @@
     <canvas class="fp-canvas" id="fp-diffraction"></canvas>
 
     <div class="fp-hairline"></div>
-    <div class="fp-date">June 18th 2026</div>
+    <div class="fp-date">June 25th 2026</div>
 
     <div class="fp-textstack">
       <div class="fp-textstack-inner">
@@ -186,7 +186,7 @@ <h1 class="fp-title">Structural Biology Has More to Teach AI Than Atomic Coordin
     </div>
 
     <div class="fp-logo">
-      <img src="{{ '/assets/posts/2026-06-18-rss-joins-omsf/rss-wordmark.png' | relative_url }}"
+      <img src="{{ '/assets/posts/2026-06-25-rss-joins-omsf/rss-wordmark.png' | relative_url }}"
            alt="Reciprocal Space Station" />
     </div>
   </div>
diff --git a/assets/posts/2026-06-18-rss-joins-omsf/rss-wordmark.png b/assets/posts/2026-06-25-rss-joins-omsf/rss-wordmark.png
similarity index 100%
rename from assets/posts/2026-06-18-rss-joins-omsf/rss-wordmark.png
rename to assets/posts/2026-06-25-rss-joins-omsf/rss-wordmark.png

From d9904e18ee9e5bdd3f38b6f794853879257d0c54 Mon Sep 17 00:00:00 2001
From: Alisia Fadini <alisia.fadini15@imperial.ac.uk>
Date: Mon, 22 Jun 2026 18:41:13 -0400
Subject: [PATCH 05/22] Enable future-dated posts so the June 25 post builds in
 previews/live

---
 _config.yml | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/_config.yml b/_config.yml
index 28b7e55..98786a3 100644
--- a/_config.yml
+++ b/_config.yml
@@ -1,5 +1,9 @@
 title: Reciprocal Space Station
 
+# Build posts with a future date (e.g. the June 25 announcement) so they
+# appear in PR previews and on the live site without waiting for the date.
+future: true
+
 plugins:
   - jekyll-font-awesome-sass
 

From f1b98d698bd33e6c5a50941973ea6c5e42ad2f50 Mon Sep 17 00:00:00 2001
From: tjlane <thomas.joseph.lane@gmail.com>
Date: Mon, 13 Jul 2026 11:09:38 -0400
Subject: [PATCH 06/22] Fill RSS-joins-OMSF post body with launch announcement
 copy
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

Replace the placeholder scaffold in the fancy-hero post with the real
launch copy: intro + open questions, RSS × OMSF, the measurements-to-
function roadmap, "Where we are going" (ensemble fitting, data velocity,
training new models), Why OMSF, and Join us. Resolve project [link]
placeholders to their repos and add the OMSF/forum/contact links.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
---
 _posts/2026-06-25-rss-joins-omsf.html | 180 +++++++++++++++++++++++---
 1 file changed, 162 insertions(+), 18 deletions(-)

diff --git a/_posts/2026-06-25-rss-joins-omsf.html b/_posts/2026-06-25-rss-joins-omsf.html
index df00796..1655dfe 100644
--- a/_posts/2026-06-25-rss-joins-omsf.html
+++ b/_posts/2026-06-25-rss-joins-omsf.html
@@ -218,24 +218,168 @@ <h1 class="fp-title">Structural Biology Has More to Teach AI Than Atomic Coordin
       <!-- Main article -->
       <div class="col-12 col-md-10 offset-md-1 col-lg-9 offset-lg-0 py-4" id="content-body">
 
-        <p class="lead reveal">Placeholder lede. The Reciprocal Space Station is joining the
-          Open Molecular Software Foundation — a short, punchy opening paragraph that sets up
-          why this matters goes here.</p>
-
-        <h2 class="reveal">Why this matters</h2>
-        <p class="reveal">Placeholder body paragraph. Replace this with the real announcement
-          copy. Each <code>.reveal</code> block fades and slides up as it scrolls into view,
-          using the site's existing IntersectionObserver.</p>
-        <p class="reveal">A second placeholder paragraph so you can see consecutive reveals
-          staggering naturally as you scroll down the page.</p>
-
-        <h3 class="reveal">A subsection</h3>
-        <p class="reveal">Subsection placeholder — note that <code>h3</code> subheadings render
-          in RSS blue, exactly like the rest of the site's posts.</p>
-
-        <h2 class="reveal">What comes next</h2>
-        <p class="reveal">Closing placeholder paragraph — links, call to action, or a quote can
-          live here once the real content is ready.</p>
+        <p class="lead reveal">The first wave of biomolecular AI learned from atomic models and
+          sequencing data. That was the right place to start: deposited structures have been the
+          common language of structural biology to date, enabling decades of drug discovery,
+          simulation and protein design. It was natural that <em>structure</em> prediction was the
+          first big breakthrough in AI-powered bioscience.</p>
+
+        <p class="reveal">But experiments do not begin with the coordinates that make up a
+          structure. The data behind the structure — every diffraction pattern, scattering event,
+          particle image — contains more information than can be captured in a single structure.
+          These data contain information about what molecules actually <em>do</em>: how they move,
+          fluctuate, bind, and respond to perturbation. To deposit coordinates, however,
+          experimentalists need to come up with the best single structure that explains most of the
+          data — treating real variability as noise. Highlighting what is lost, current structure
+          predictors struggle with dynamics in part because they learned from this compressed
+          information. They trained on the world as it was deposited, not as it was measured.</p>
+
+        <p class="reveal">Those lost dynamics hold the key to some of the most important open
+          questions in molecular biology:</p>
+
+        <ul class="reveal">
+          <li><em>When is a mutation harmless, and when does it reshape a functional state?</em></li>
+          <li><em>When does a cryptic pocket open, and how can we target it?</em></li>
+          <li><em>How do proteins transmit information across distance?</em></li>
+          <li><em>How do molecular structure, dynamics, and function change with cellular context?</em></li>
+        </ul>
+
+        <p class="reveal">The answers — dynamics — appear in the data we collect today and will be
+          the central pursuit of the experiments of tomorrow. To capture them, we need to develop
+          the next generation of models, beyond single structures, by returning to the measurements
+          themselves. We need the computational infrastructure to make experimental data usable for
+          inference of dynamics today and enable training of the foundation models of tomorrow.</p>
+
+        <h2 class="reveal">RSS &times; OMSF</h2>
+
+        <p class="reveal">In pursuit of this future, we are excited to announce that the Reciprocal
+          Space Station, or RSS, is joining the Open Molecular Software Foundation
+          (<a href="https://omsf.io">OMSF</a>) to tackle that goal together: building open,
+          production-grade software that turns structural biology experiments into dynamics and
+          function. OMSF's help with governance, administration, software engineering, and
+          everything open source will enable RSS to expand our mission and tackle the frontier of
+          dynamic structural software.</p>
+
+        <p class="reveal">But what is a Reciprocal Space Station anyways? Our cheeky name riffs off
+          reciprocal space, the mathematical representation where many structural biology
+          experiments naturally live. Atomic-resolution structural biology relies on scattering:
+          X-rays, electrons, or neutrons bounce off a sample, and a detector records the information
+          imprinted on that radiation. Due to that radiation's wave-like nature, the process is
+          naturally described in reciprocal space, the Fourier-space world where crystallography,
+          cryo-EM, diffuse scattering, and small-angle methods meet. We are setting out to explore
+          this data space from our home base, the Reciprocal Space Station.</p>
+
+        <p class="reveal">RSS is part of the larger structural biology community, and is committed to
+          supporting and sustaining that community and ongoing structural software development. Stay
+          tuned, as more details about how RSS plans to operate and contribute will follow in a
+          future post!</p>
+
+        <h2 class="reveal">Building a path from measurements to function</h2>
+
+        <p class="reveal">The bottleneck RSS is tackling is not just one missing algorithm, but a
+          stack of connected problems. Experimental data needs a translation layer: from
+          facility-specific files and scientific conventions to context-aware arrays and tensors.
+          Forward models, which map conformations back to experimental observables, need to be
+          accurate, differentiable, and fast. Finally, we need the ability to go beyond single
+          structures to conformational distributions that change depending on environment,
+          composition, and time.</p>
+
+        <p class="reveal">Our projects target that stack directly:
+          <a href="https://github.com/rs-station/reciprocalspaceship">reciprocalspaceship</a> for
+          programmable crystallographic data;
+          <a href="https://github.com/rs-station/careless">Careless</a>,
+          <a href="https://github.com/rs-station/meteor">Meteor</a>, and
+          <a href="https://github.com/rs-station/laue-dials">Laue-DIALS</a> for difficult
+          diffraction and time-resolved regimes;
+          <a href="https://github.com/Hekstra-Lab/SFcalculator_torch">SFCalculator</a> for
+          differentiable forward modeling; and
+          <a href="https://github.com/rs-station/rocket">ROCKET</a> and
+          <a href="https://arxiv.org/abs/2602.05285">EmbedOpt</a> for observable-guided structure
+          and ensemble inference.</p>
+
+        <p class="reveal">Together, these efforts define the RSS roadmap: make structural biology
+          data accessible and interoperable, implement the underlying physics as differentiable
+          models, and train the next generation of biomolecular AI on data, not models.</p>
+
+        <p class="reveal">Concretely, RSS is building three connected layers.</p>
+
+        <p class="reveal">First, data libraries that make diffraction, scattering, imaging, and
+          time-resolved data easier to represent, manipulate, validate, and share. These are
+          building blocks that experimentalists, software developers, and model builders can use and
+          reuse.</p>
+
+        <p class="reveal">Second, experimental forward models that connect molecular structures and
+          ensembles to the measurements instruments actually record. These models need to be
+          physically correct, noise-aware, differentiable, performant, and usable across modalities
+          including crystallography, cryo-EM, SAXS, diffuse scattering, and beyond.</p>
+
+        <p class="reveal">Third, modeling systems that use experimental data directly: not only to
+          validate a final structure, but to guide structure determination, ensemble fitting, and
+          model training.</p>
+
+        <h2 class="reveal">Where we are going</h2>
+
+        <p class="reveal">We've already started down this path, with some exciting progress.</p>
+
+        <h3 class="reveal">Ensemble fitting from raw observables</h3>
+        <p class="reveal">In dynamic regions, building a structural model by hand is often
+          impossible. To address this challenge, we are developing engines that automatically fit
+          distributions of conformations directly to experimental data. In a first application,
+          without human intervention our system fit multiple conformers to crystallographic
+          measurements taken across a range of temperatures, yielding better models (per
+          cross-validation) than the best single-structure model. Crucially, this let us capture
+          allostery, where changes on one side of a protein propagate to another, spatially removed
+          site. Understanding allostery enables new opportunities in drug discovery and is
+          impossible to infer from a single-structure model. Next, we are extending this machinery
+          to ligands, waters, nucleic acids, and larger molecular assemblies, to understand the
+          function and dynamics of these critical cellular actors.</p>
+
+        <h3 class="reveal">Data velocity for modern instruments</h3>
+        <p class="reveal">Once interpretation no longer requires hours of manual model building per
+          dataset, scale becomes the opportunity. We are partnering with efforts such as OpenBind
+          and OpenADMET to help turn modern high-throughput structural experiments into usable data
+          streams for cofolding, binding, and ADMET modeling.</p>
+
+        <h3 class="reveal">From guiding existing models to training new ones</h3>
+        <p class="reveal">Alongside sister projects such as OpenFold, we aim to close the
+          experiment-model loop: measurements guide models, while models help design and interpret
+          experiments. We plan to build the infrastructure that will enable the next big
+          breakthroughs in biomolecular AI: cofolding, protein-protein interface prediction, and the
+          design of protein dynamics.</p>
+
+        <h2 class="reveal">Why OMSF</h2>
+
+        <p class="reveal">Infrastructure requires planning, maintenance, and resources. We'll need
+          to collaborate across structural biology, computation, simulation, drug discovery, and
+          experimental facilities. To succeed, our community will need to produce and support code
+          that is not only scientifically impactful, but available, reliable, extensible, and
+          community-driven.</p>
+
+        <p class="reveal">That is why RSS is joining the Open Molecular Software Foundation to
+          further this mission in service of the structural biology and AI communities. OMSF
+          provides a home for open molecular software projects that support entire fields. Alongside
+          consortia such as OpenFold, Open Force Field, OpenADMET, and OpenBind, RSS will help build
+          the open infrastructure needed for the next generation of structural biology.</p>
+
+        <p class="reveal">To realize the next generation of biomolecular AI, scientific excellence
+          is necessary but not sufficient. Automation, performance, usability, documentation,
+          testing, education, and long-term stewardship are first-order concerns of RSS, alongside
+          scientific insight.</p>
+
+        <h2 class="reveal">Join us</h2>
+
+        <p class="reveal">At its heart, RSS is an open community of reciprocal astronauts:
+          structural biologists, software developers, experimentalists, method builders, and
+          partners who believe experiments have more to teach us than a single static model.</p>
+
+        <p class="reveal">If you are excited about open software, experimental data, molecular
+          dynamics, and the future of structural biology, reach out at
+          <a href="mailto:crew@rs-station.org">crew@rs-station.org</a>, find us on
+          <a href="https://github.com/rs-station">GitHub</a>, or join our
+          <a href="https://discourse.rs-station.org">community forum</a>.</p>
+
+        <p class="reveal">Structural biology has more to teach AI than coordinates. Let's build the
+          tools to help it speak.</p>
 
         <p class="reveal mt-5"><a href="{{ '/blog.html' | relative_url }}">⬅️ Back to blog posts</a></p>
 

From b9044f7e35e11ccbb7194e4dab110329335f9a82 Mon Sep 17 00:00:00 2001
From: tjlane <thomas.joseph.lane@gmail.com>
Date: Mon, 13 Jul 2026 11:40:12 -0400
Subject: [PATCH 07/22] Point EmbedOpt at its repo; link
 OpenBind/OpenADMET/OpenFold

Swap the EmbedOpt arXiv link for github.com/rs-station/embedopt, and
link the first mention of each partner consortium: OpenBind (openbind.uk)
and OpenADMET/OpenFold (their OMSF project pages).

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
---
 _posts/2026-06-25-rss-joins-omsf.html | 13 ++++++++-----
 1 file changed, 8 insertions(+), 5 deletions(-)

diff --git a/_posts/2026-06-25-rss-joins-omsf.html b/_posts/2026-06-25-rss-joins-omsf.html
index 1655dfe..b8e4a7b 100644
--- a/_posts/2026-06-25-rss-joins-omsf.html
+++ b/_posts/2026-06-25-rss-joins-omsf.html
@@ -294,7 +294,7 @@ <h2 class="reveal">Building a path from measurements to function</h2>
           <a href="https://github.com/Hekstra-Lab/SFcalculator_torch">SFCalculator</a> for
           differentiable forward modeling; and
           <a href="https://github.com/rs-station/rocket">ROCKET</a> and
-          <a href="https://arxiv.org/abs/2602.05285">EmbedOpt</a> for observable-guided structure
+          <a href="https://github.com/rs-station/embedopt">EmbedOpt</a> for observable-guided structure
           and ensemble inference.</p>
 
         <p class="reveal">Together, these efforts define the RSS roadmap: make structural biology
@@ -336,12 +336,15 @@ <h3 class="reveal">Ensemble fitting from raw observables</h3>
 
         <h3 class="reveal">Data velocity for modern instruments</h3>
         <p class="reveal">Once interpretation no longer requires hours of manual model building per
-          dataset, scale becomes the opportunity. We are partnering with efforts such as OpenBind
-          and OpenADMET to help turn modern high-throughput structural experiments into usable data
-          streams for cofolding, binding, and ADMET modeling.</p>
+          dataset, scale becomes the opportunity. We are partnering with efforts such as
+          <a href="https://openbind.uk">OpenBind</a> and
+          <a href="https://omsf.io/programs/projects/openadmet/">OpenADMET</a> to help turn modern
+          high-throughput structural experiments into usable data streams for cofolding, binding,
+          and ADMET modeling.</p>
 
         <h3 class="reveal">From guiding existing models to training new ones</h3>
-        <p class="reveal">Alongside sister projects such as OpenFold, we aim to close the
+        <p class="reveal">Alongside sister projects such as
+          <a href="https://omsf.io/programs/projects/openfold/">OpenFold</a>, we aim to close the
           experiment-model loop: measurements guide models, while models help design and interpret
           experiments. We plan to build the infrastructure that will enable the next big
           breakthroughs in biomolecular AI: cofolding, protein-protein interface prediction, and the

From 8a60331234697539f6b774515a2ff61e2772dffd Mon Sep 17 00:00:00 2001
From: tjlane <thomas.joseph.lane@gmail.com>
Date: Mon, 13 Jul 2026 11:41:49 -0400
Subject: [PATCH 08/22] Link OpenFold/OpenADMET to their own sites

Use openfold.io and openadmet.org instead of the OMSF project pages.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
---
 _posts/2026-06-25-rss-joins-omsf.html | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/_posts/2026-06-25-rss-joins-omsf.html b/_posts/2026-06-25-rss-joins-omsf.html
index b8e4a7b..0234c48 100644
--- a/_posts/2026-06-25-rss-joins-omsf.html
+++ b/_posts/2026-06-25-rss-joins-omsf.html
@@ -338,13 +338,13 @@ <h3 class="reveal">Data velocity for modern instruments</h3>
         <p class="reveal">Once interpretation no longer requires hours of manual model building per
           dataset, scale becomes the opportunity. We are partnering with efforts such as
           <a href="https://openbind.uk">OpenBind</a> and
-          <a href="https://omsf.io/programs/projects/openadmet/">OpenADMET</a> to help turn modern
+          <a href="https://openadmet.org/">OpenADMET</a> to help turn modern
           high-throughput structural experiments into usable data streams for cofolding, binding,
           and ADMET modeling.</p>
 
         <h3 class="reveal">From guiding existing models to training new ones</h3>
         <p class="reveal">Alongside sister projects such as
-          <a href="https://omsf.io/programs/projects/openfold/">OpenFold</a>, we aim to close the
+          <a href="https://openfold.io/">OpenFold</a>, we aim to close the
           experiment-model loop: measurements guide models, while models help design and interpret
           experiments. We plan to build the infrastructure that will enable the next big
           breakthroughs in biomolecular AI: cofolding, protein-protein interface prediction, and the

From 6275405d391d64001acdb6f879bf25765cde600f Mon Sep 17 00:00:00 2001
From: tjlane <thomas.joseph.lane@gmail.com>
Date: Mon, 13 Jul 2026 11:42:54 -0400
Subject: [PATCH 09/22] Move coordinates sentence into lead, lightly bold it

Pull "But experiments do not begin with the coordinates that make up a
structure." up as the semibold closing line of the opening paragraph.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
---
 _posts/2026-06-25-rss-joins-omsf.html | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)

diff --git a/_posts/2026-06-25-rss-joins-omsf.html b/_posts/2026-06-25-rss-joins-omsf.html
index 0234c48..3c72ffd 100644
--- a/_posts/2026-06-25-rss-joins-omsf.html
+++ b/_posts/2026-06-25-rss-joins-omsf.html
@@ -222,10 +222,11 @@ <h1 class="fp-title">Structural Biology Has More to Teach AI Than Atomic Coordin
           sequencing data. That was the right place to start: deposited structures have been the
           common language of structural biology to date, enabling decades of drug discovery,
           simulation and protein design. It was natural that <em>structure</em> prediction was the
-          first big breakthrough in AI-powered bioscience.</p>
+          first big breakthrough in AI-powered bioscience.
+          <strong style="font-weight: 600;">But experiments do not begin with the coordinates that
+          make up a structure.</strong></p>
 
-        <p class="reveal">But experiments do not begin with the coordinates that make up a
-          structure. The data behind the structure — every diffraction pattern, scattering event,
+        <p class="reveal">The data behind the structure — every diffraction pattern, scattering event,
           particle image — contains more information than can be captured in a single structure.
           These data contain information about what molecules actually <em>do</em>: how they move,
           fluctuate, bind, and respond to perturbation. To deposit coordinates, however,

From d5d1e470395da5d6d999b66a408d988713eca72b Mon Sep 17 00:00:00 2001
From: tjlane <thomas.joseph.lane@gmail.com>
Date: Mon, 13 Jul 2026 11:50:41 -0400
Subject: [PATCH 10/22] Link every consortium mention; add OpenForceField

Link all instances of OpenFold/OpenADMET/OpenBind (not just the first),
and link Open Force Field -> OpenForceField (openforcefield.org) in the
Why OMSF recap.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
---
 _posts/2026-06-25-rss-joins-omsf.html | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)

diff --git a/_posts/2026-06-25-rss-joins-omsf.html b/_posts/2026-06-25-rss-joins-omsf.html
index 3c72ffd..30f12e7 100644
--- a/_posts/2026-06-25-rss-joins-omsf.html
+++ b/_posts/2026-06-25-rss-joins-omsf.html
@@ -362,7 +362,10 @@ <h2 class="reveal">Why OMSF</h2>
         <p class="reveal">That is why RSS is joining the Open Molecular Software Foundation to
           further this mission in service of the structural biology and AI communities. OMSF
           provides a home for open molecular software projects that support entire fields. Alongside
-          consortia such as OpenFold, Open Force Field, OpenADMET, and OpenBind, RSS will help build
+          consortia such as <a href="https://openfold.io/">OpenFold</a>,
+          <a href="https://openforcefield.org/">OpenForceField</a>,
+          <a href="https://openadmet.org/">OpenADMET</a>, and
+          <a href="https://openbind.uk">OpenBind</a>, RSS will help build
           the open infrastructure needed for the next generation of structural biology.</p>
 
         <p class="reveal">To realize the next generation of biomolecular AI, scientific excellence

From 9adf5e528ef627695e8012ebbdb0c22cfb283b9b Mon Sep 17 00:00:00 2001
From: tjlane <thomas.joseph.lane@gmail.com>
Date: Mon, 13 Jul 2026 13:21:11 -0400
Subject: [PATCH 11/22] final minor edits, tweaks

---
 _posts/2026-06-25-rss-joins-omsf.html | 14 +++++++-------
 1 file changed, 7 insertions(+), 7 deletions(-)

diff --git a/_posts/2026-06-25-rss-joins-omsf.html b/_posts/2026-06-25-rss-joins-omsf.html
index 30f12e7..c2f7129 100644
--- a/_posts/2026-06-25-rss-joins-omsf.html
+++ b/_posts/2026-06-25-rss-joins-omsf.html
@@ -227,11 +227,11 @@ <h1 class="fp-title">Structural Biology Has More to Teach AI Than Atomic Coordin
           make up a structure.</strong></p>
 
         <p class="reveal">The data behind the structure — every diffraction pattern, scattering event,
-          particle image — contains more information than can be captured in a single structure.
+          particle image — contains more information than can be captured in a single conformation.
           These data contain information about what molecules actually <em>do</em>: how they move,
           fluctuate, bind, and respond to perturbation. To deposit coordinates, however,
-          experimentalists need to come up with the best single structure that explains most of the
-          data — treating real variability as noise. Highlighting what is lost, current structure
+          experimentalists need to come up with the best single snapshot that explains most of the
+          data — treating real variability as noise. Tellingly, current structure
           predictors struggle with dynamics in part because they learned from this compressed
           information. They trained on the world as it was deposited, not as it was measured.</p>
 
@@ -261,7 +261,7 @@ <h2 class="reveal">RSS &times; OMSF</h2>
           everything open source will enable RSS to expand our mission and tackle the frontier of
           dynamic structural software.</p>
 
-        <p class="reveal">But what is a Reciprocal Space Station anyways? Our cheeky name riffs off
+        <p class="reveal">But what is a Reciprocal Space Station anyway? Our cheeky name riffs off
           reciprocal space, the mathematical representation where many structural biology
           experiments naturally live. Atomic-resolution structural biology relies on scattering:
           X-rays, electrons, or neutrons bounce off a sample, and a detector records the information
@@ -271,7 +271,7 @@ <h2 class="reveal">RSS &times; OMSF</h2>
           this data space from our home base, the Reciprocal Space Station.</p>
 
         <p class="reveal">RSS is part of the larger structural biology community, and is committed to
-          supporting and sustaining that community and ongoing structural software development. Stay
+          supporting and sustaining it and ongoing structural software development. Stay
           tuned, as more details about how RSS plans to operate and contribute will follow in a
           future post!</p>
 
@@ -326,11 +326,11 @@ <h3 class="reveal">Ensemble fitting from raw observables</h3>
         <p class="reveal">In dynamic regions, building a structural model by hand is often
           impossible. To address this challenge, we are developing engines that automatically fit
           distributions of conformations directly to experimental data. In a first application,
-          without human intervention our system fit multiple conformers to crystallographic
+          without human intervention, our system fit multiple conformers to crystallographic
           measurements taken across a range of temperatures, yielding better models (per
           cross-validation) than the best single-structure model. Crucially, this let us capture
           allostery, where changes on one side of a protein propagate to another, spatially removed
-          site. Understanding allostery enables new opportunities in drug discovery and is
+          site. Allostery enables new opportunities in drug discovery, but is
           impossible to infer from a single-structure model. Next, we are extending this machinery
           to ligands, waters, nucleic acids, and larger molecular assemblies, to understand the
           function and dynamics of these critical cellular actors.</p>

From 86faf3732c46cc9d8d006754eebf2723d109c1d9 Mon Sep 17 00:00:00 2001
From: tjlane <thomas.joseph.lane@gmail.com>
Date: Mon, 13 Jul 2026 14:28:09 -0400
Subject: [PATCH 12/22] alisia fix for no data

---
 _posts/2026-06-25-rss-joins-omsf.html | 20 +++++++++-----------
 1 file changed, 9 insertions(+), 11 deletions(-)

diff --git a/_posts/2026-06-25-rss-joins-omsf.html b/_posts/2026-06-25-rss-joins-omsf.html
index c2f7129..15e87f1 100644
--- a/_posts/2026-06-25-rss-joins-omsf.html
+++ b/_posts/2026-06-25-rss-joins-omsf.html
@@ -323,17 +323,15 @@ <h2 class="reveal">Where we are going</h2>
         <p class="reveal">We've already started down this path, with some exciting progress.</p>
 
         <h3 class="reveal">Ensemble fitting from raw observables</h3>
-        <p class="reveal">In dynamic regions, building a structural model by hand is often
-          impossible. To address this challenge, we are developing engines that automatically fit
-          distributions of conformations directly to experimental data. In a first application,
-          without human intervention, our system fit multiple conformers to crystallographic
-          measurements taken across a range of temperatures, yielding better models (per
-          cross-validation) than the best single-structure model. Crucially, this let us capture
-          allostery, where changes on one side of a protein propagate to another, spatially removed
-          site. Allostery enables new opportunities in drug discovery, but is
-          impossible to infer from a single-structure model. Next, we are extending this machinery
-          to ligands, waters, nucleic acids, and larger molecular assemblies, to understand the
-          function and dynamics of these critical cellular actors.</p>
+        <p class="reveal">In dynamic regions, building a 
+          structural model by hand is often impossible. To address this challenge, we are developing
+          engines that automatically fit distributions of conformations directly to experimental 
+          data.  This will enable us to capture allostery, where changes on one side of a protein 
+          propagate to another, spatially removed site. Understanding allostery enables new 
+          opportunities in drug discovery and is impossible to infer from a single-structure model. 
+          We are therefore extending our machinery to ensembles (stay tuned for an upcoming post), 
+          as well as ligands, waters, nucleic acids, and larger molecular assemblies, to understand
+          the function and dynamics of these critical cellular actors.</p>
 
         <h3 class="reveal">Data velocity for modern instruments</h3>
         <p class="reveal">Once interpretation no longer requires hours of manual model building per

From a9f25a1ecac1262607205b47f1713f6c16a85a5d Mon Sep 17 00:00:00 2001
From: tjlane <thomas.joseph.lane@gmail.com>
Date: Mon, 13 Jul 2026 14:33:18 -0400
Subject: [PATCH 13/22] fix today tomorrow

---
 _posts/2026-06-25-rss-joins-omsf.html | 9 +++++----
 1 file changed, 5 insertions(+), 4 deletions(-)

diff --git a/_posts/2026-06-25-rss-joins-omsf.html b/_posts/2026-06-25-rss-joins-omsf.html
index 15e87f1..2e52ec0 100644
--- a/_posts/2026-06-25-rss-joins-omsf.html
+++ b/_posts/2026-06-25-rss-joins-omsf.html
@@ -247,9 +247,10 @@ <h1 class="fp-title">Structural Biology Has More to Teach AI Than Atomic Coordin
 
         <p class="reveal">The answers — dynamics — appear in the data we collect today and will be
           the central pursuit of the experiments of tomorrow. To capture them, we need to develop
-          the next generation of models, beyond single structures, by returning to the measurements
-          themselves. We need the computational infrastructure to make experimental data usable for
-          inference of dynamics today and enable training of the foundation models of tomorrow.</p>
+          models that go beyond single structures and can represent distributions. This will require
+          a return to the measurements themselves. We need to build the computational infrastructure that will make
+          experimental data usable for inference of dynamics from experiment and enable training of 
+          foundation models based on distributions of structure.</p>
 
         <h2 class="reveal">RSS &times; OMSF</h2>
 
@@ -300,7 +301,7 @@ <h2 class="reveal">Building a path from measurements to function</h2>
 
         <p class="reveal">Together, these efforts define the RSS roadmap: make structural biology
           data accessible and interoperable, implement the underlying physics as differentiable
-          models, and train the next generation of biomolecular AI on data, not models.</p>
+          models, and train the next generation of biomolecular AI on data, not structures.</p>
 
         <p class="reveal">Concretely, RSS is building three connected layers.</p>
 

From 68939d8fa5b5323284ff356c34bef7b81d07fad6 Mon Sep 17 00:00:00 2001
From: tjlane <thomas.joseph.lane@gmail.com>
Date: Mon, 13 Jul 2026 14:34:47 -0400
Subject: [PATCH 14/22] alisia new version

---
 _posts/2026-06-25-rss-joins-omsf.html | 10 +---------
 1 file changed, 1 insertion(+), 9 deletions(-)

diff --git a/_posts/2026-06-25-rss-joins-omsf.html b/_posts/2026-06-25-rss-joins-omsf.html
index 2e52ec0..09276e0 100644
--- a/_posts/2026-06-25-rss-joins-omsf.html
+++ b/_posts/2026-06-25-rss-joins-omsf.html
@@ -324,15 +324,7 @@ <h2 class="reveal">Where we are going</h2>
         <p class="reveal">We've already started down this path, with some exciting progress.</p>
 
         <h3 class="reveal">Ensemble fitting from raw observables</h3>
-        <p class="reveal">In dynamic regions, building a 
-          structural model by hand is often impossible. To address this challenge, we are developing
-          engines that automatically fit distributions of conformations directly to experimental 
-          data.  This will enable us to capture allostery, where changes on one side of a protein 
-          propagate to another, spatially removed site. Understanding allostery enables new 
-          opportunities in drug discovery and is impossible to infer from a single-structure model. 
-          We are therefore extending our machinery to ensembles (stay tuned for an upcoming post), 
-          as well as ligands, waters, nucleic acids, and larger molecular assemblies, to understand
-          the function and dynamics of these critical cellular actors.</p>
+        <p class="reveal">In dynamic regions, building a structural model by hand is often impossible. To address this challenge, we are developing engines that automatically fit distributions of conformations directly to experimental data.  This will enable us to capture allostery, where changes on one side of a protein propagate to another, spatially removed site. Understanding allostery enables new opportunities in drug discovery and is impossible to infer from a single-structure model. We are therefore extending our machinery to ensembles, stay tuned for an upcoming post!</p>
 
         <h3 class="reveal">Data velocity for modern instruments</h3>
         <p class="reveal">Once interpretation no longer requires hours of manual model building per

From b1a4c74ec44383d23c05065ac33b19521b5b93e1 Mon Sep 17 00:00:00 2001
From: tjlane <thomas.joseph.lane@gmail.com>
Date: Mon, 13 Jul 2026 14:42:16 -0400
Subject: [PATCH 15/22] last fixes

---
 _posts/2026-06-25-rss-joins-omsf.html | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/_posts/2026-06-25-rss-joins-omsf.html b/_posts/2026-06-25-rss-joins-omsf.html
index 09276e0..86b6c4e 100644
--- a/_posts/2026-06-25-rss-joins-omsf.html
+++ b/_posts/2026-06-25-rss-joins-omsf.html
@@ -249,7 +249,7 @@ <h1 class="fp-title">Structural Biology Has More to Teach AI Than Atomic Coordin
           the central pursuit of the experiments of tomorrow. To capture them, we need to develop
           models that go beyond single structures and can represent distributions. This will require
           a return to the measurements themselves. We need to build the computational infrastructure that will make
-          experimental data usable for inference of dynamics from experiment and enable training of 
+          experimental data usable for inference of dynamics and enable training of
           foundation models based on distributions of structure.</p>
 
         <h2 class="reveal">RSS &times; OMSF</h2>
@@ -324,7 +324,7 @@ <h2 class="reveal">Where we are going</h2>
         <p class="reveal">We've already started down this path, with some exciting progress.</p>
 
         <h3 class="reveal">Ensemble fitting from raw observables</h3>
-        <p class="reveal">In dynamic regions, building a structural model by hand is often impossible. To address this challenge, we are developing engines that automatically fit distributions of conformations directly to experimental data.  This will enable us to capture allostery, where changes on one side of a protein propagate to another, spatially removed site. Understanding allostery enables new opportunities in drug discovery and is impossible to infer from a single-structure model. We are therefore extending our machinery to ensembles, stay tuned for an upcoming post!</p>
+        <p class="reveal">In dynamic regions, building a structural model by hand is often impossible. To address this challenge, we are developing engines that automatically fit distributions of conformations directly to experimental data. This will enable us to capture allostery, where changes on one side of a protein propagate to another, spatially removed site. Allostery opens new opportunities in drug discovery and is impossible to infer from a single-structure model. We are therefore extending our machinery to ensembles — stay tuned for an upcoming post!</p>
 
         <h3 class="reveal">Data velocity for modern instruments</h3>
         <p class="reveal">Once interpretation no longer requires hours of manual model building per
@@ -359,7 +359,7 @@ <h2 class="reveal">Why OMSF</h2>
           <a href="https://openbind.uk">OpenBind</a>, RSS will help build
           the open infrastructure needed for the next generation of structural biology.</p>
 
-        <p class="reveal">To realize the next generation of biomolecular AI, scientific excellence
+        <p class="reveal">To realize this mission, scientific excellence
           is necessary but not sufficient. Automation, performance, usability, documentation,
           testing, education, and long-term stewardship are first-order concerns of RSS, alongside
           scientific insight.</p>

From 776eb34a77139d3c0b8aef2e0407afdea1630b72 Mon Sep 17 00:00:00 2001
From: tjlane <thomas.joseph.lane@gmail.com>
Date: Tue, 14 Jul 2026 09:58:21 -0400
Subject: [PATCH 16/22] Copyedit body and fix TOC anchor scroll offset

Light copyedits (drop redundant "from experiment", fix a double space
and comma splice, avoid enable/allostery and next-generation repeats),
and add scroll-margin-top to #content-body headings so TOC/anchor jumps
land the section title just below the fixed navbar instead of under it.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
---
 _posts/2026-06-25-rss-joins-omsf.html | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/_posts/2026-06-25-rss-joins-omsf.html b/_posts/2026-06-25-rss-joins-omsf.html
index 86b6c4e..4f6d95c 100644
--- a/_posts/2026-06-25-rss-joins-omsf.html
+++ b/_posts/2026-06-25-rss-joins-omsf.html
@@ -150,6 +150,10 @@
   .fp-after { background: #ffffff; }
   /* RSS-blue section titles for this announcement (site default h2 is near-black) */
   body.fp-hide-nav #content-body > h2 { color: var(--rss-blue); }
+  /* Offset anchor jumps (TOC clicks) so the section title clears the fixed
+     navbar and sits at the top of the view, rather than under it. */
+  #content-body h2,
+  #content-body h3 { scroll-margin-top: 80px; }
 
   @media (max-width: 991.98px) {
     .fp-subtitle { display: none; } /* keep the small-screen hero uncluttered */

From 6a937fae8e18af8a70188e92e3abffcaa85168a9 Mon Sep 17 00:00:00 2001
From: tjlane <thomas.joseph.lane@gmail.com>
Date: Fri, 17 Jul 2026 12:49:08 -0400
Subject: [PATCH 17/22] update

---
 _posts/2026-06-25-rss-joins-omsf.html | 49 ++++++++++++++++++++++-----
 1 file changed, 40 insertions(+), 9 deletions(-)

diff --git a/_posts/2026-06-25-rss-joins-omsf.html b/_posts/2026-06-25-rss-joins-omsf.html
index 4f6d95c..2435e5b 100644
--- a/_posts/2026-06-25-rss-joins-omsf.html
+++ b/_posts/2026-06-25-rss-joins-omsf.html
@@ -236,24 +236,55 @@ <h1 class="fp-title">Structural Biology Has More to Teach AI Than Atomic Coordin
           fluctuate, bind, and respond to perturbation. To deposit coordinates, however,
           experimentalists need to come up with the best single snapshot that explains most of the
           data — treating real variability as noise. Tellingly, current structure
-          predictors struggle with dynamics in part because they learned from this compressed
+          predictors struggle with structural heterogeneity in part because they learned from this compressed
           information. They trained on the world as it was deposited, not as it was measured.</p>
 
-        <p class="reveal">Those lost dynamics hold the key to some of the most important open
+        <p class="reveal">That lost information holds the key to some of the most important open
           questions in molecular biology:</p>
 
         <ul class="reveal">
           <li><em>When is a mutation harmless, and when does it reshape a functional state?</em></li>
           <li><em>When does a cryptic pocket open, and how can we target it?</em></li>
-          <li><em>How do proteins transmit information across distance?</em></li>
+          <li><em>How do proteins transmit and transform information through their structures?</em></li>
+          <li><em>Can we understand the details of how water interacts with proteins — and use that to increase ligand potency?</em></li>
           <li><em>How do molecular structure, dynamics, and function change with cellular context?</em></li>
         </ul>
 
-        <p class="reveal">The answers — dynamics — appear in the data we collect today and will be
-          the central pursuit of the experiments of tomorrow. To capture them, we need to develop
-          models that go beyond single structures and can represent distributions. This will require
-          a return to the measurements themselves. We need to build the computational infrastructure that will make
-          experimental data usable for inference of dynamics and enable training of
+        <p class="reveal">The answers are hidden in the data we collect today and will be
+          the central pursuit of the experiments of tomorrow. </p>
+            
+        <p class="reveal">To extract them, we need to develop
+          models that go beyond the static coordinates that we use to represent structure today. 
+          These models are hand-fit to experimental data, built by humans, for humans. They have been 
+          incredibly successful: the Protein Data Bank (PDB) contains &gt;250k structures that produced
+          9 Nobel prizes, guided &gt;80% of approved drugs, and trained AlphaFold, hailed as the first
+          scientific breakthrough powered by AI. But lists of coordinates also limit the ability to
+          reason about essential but complex aspects of structure: allostery, dynamics, water networks, 
+          partially ordered and disordered regions — in short, heterogeneity. When structure 
+          is compressed into a single coordinate model, this information is lost.
+          </p>
+
+        <p class="reveal">We see a future in which AI models enable us to recapture this information,
+          tackle the complexity of heterogeneity, and thereby fundamentally change how structure 
+          impacts biology and medicine. Our vision is to ultimately bypass the coordinate models we
+          use currently and instead train AI models directly on structural experiments. In doing so, we
+          can leverage machine intelligence to do science beyond the limits of humans and coordinates.
+        </p>
+
+        <p class="reveal">To see this future more concretely, take a practical problem that
+          faces drug hunters today: developing a molecular glue that creates a new
+          protein-protein interface between two proteins. This 
+          task is essentially intractable for current rational, structure-based drug design. However, 
+          with the assistance of a biomolecular foundation model capable of abstractly reasoning 
+          about how small molecule binding induces structural changes, what drives protein-protein 
+          interactions, and the conformational preferences of any final complex, this task could 
+          become routine.
+        </p>
+          
+        <p class="reveal">Surpassing the limits of static coordinates to realize this vision will require 
+          a return to the measurements themselves. We need to build the AI-structure interface,
+          the computational infrastructure that will make
+          experimental data usable for inference of structural heterogeneity and enable training of
           foundation models based on distributions of structure.</p>
 
         <h2 class="reveal">RSS &times; OMSF</h2>
@@ -381,7 +412,7 @@ <h2 class="reveal">Join us</h2>
           <a href="https://discourse.rs-station.org">community forum</a>.</p>
 
         <p class="reveal">Structural biology has more to teach AI than coordinates. Let's build the
-          tools to help it speak.</p>
+          tools to let it speak.</p>
 
         <p class="reveal mt-5"><a href="{{ '/blog.html' | relative_url }}">⬅️ Back to blog posts</a></p>
 

From 15d744fd5a1d0f6d6da3e3ce3292b52784b84c88 Mon Sep 17 00:00:00 2001
From: tjlane <thomas.joseph.lane@gmail.com>
Date: Fri, 17 Jul 2026 12:50:17 -0400
Subject: [PATCH 18/22] consistent font size

---
 _posts/2026-06-25-rss-joins-omsf.html | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/_posts/2026-06-25-rss-joins-omsf.html b/_posts/2026-06-25-rss-joins-omsf.html
index 2435e5b..b3e24f8 100644
--- a/_posts/2026-06-25-rss-joins-omsf.html
+++ b/_posts/2026-06-25-rss-joins-omsf.html
@@ -222,7 +222,7 @@ <h1 class="fp-title">Structural Biology Has More to Teach AI Than Atomic Coordin
       <!-- Main article -->
       <div class="col-12 col-md-10 offset-md-1 col-lg-9 offset-lg-0 py-4" id="content-body">
 
-        <p class="lead reveal">The first wave of biomolecular AI learned from atomic models and
+        <p class="reveal">The first wave of biomolecular AI learned from atomic models and
           sequencing data. That was the right place to start: deposited structures have been the
           common language of structural biology to date, enabling decades of drug discovery,
           simulation and protein design. It was natural that <em>structure</em> prediction was the

From ddd26a6a8883839238f70b2479d73f1ccaaa7de3 Mon Sep 17 00:00:00 2001
From: tjlane <thomas.joseph.lane@gmail.com>
Date: Fri, 17 Jul 2026 13:05:14 -0400
Subject: [PATCH 19/22] minor tweaks

---
 _posts/2026-06-25-rss-joins-omsf.html | 25 ++++++++++++-------------
 1 file changed, 12 insertions(+), 13 deletions(-)

diff --git a/_posts/2026-06-25-rss-joins-omsf.html b/_posts/2026-06-25-rss-joins-omsf.html
index b3e24f8..3c52d05 100644
--- a/_posts/2026-06-25-rss-joins-omsf.html
+++ b/_posts/2026-06-25-rss-joins-omsf.html
@@ -234,9 +234,9 @@ <h1 class="fp-title">Structural Biology Has More to Teach AI Than Atomic Coordin
           particle image — contains more information than can be captured in a single conformation.
           These data contain information about what molecules actually <em>do</em>: how they move,
           fluctuate, bind, and respond to perturbation. To deposit coordinates, however,
-          experimentalists need to come up with the best single snapshot that explains most of the
-          data — treating real variability as noise. Tellingly, current structure
-          predictors struggle with structural heterogeneity in part because they learned from this compressed
+          experimentalists need to come up with the best single snapshot that explains the
+          data, treating real variability as noise. Tellingly, current structure
+          predictors struggle with experimental heterogeneity in part because they learned from this compressed
           information. They trained on the world as it was deposited, not as it was measured.</p>
 
         <p class="reveal">That lost information holds the key to some of the most important open
@@ -244,36 +244,35 @@ <h1 class="fp-title">Structural Biology Has More to Teach AI Than Atomic Coordin
 
         <ul class="reveal">
           <li><em>When is a mutation harmless, and when does it reshape a functional state?</em></li>
-          <li><em>When does a cryptic pocket open, and how can we target it?</em></li>
+          <li><em>Where are cryptic pockets, and how can we target them?</em></li>
           <li><em>How do proteins transmit and transform information through their structures?</em></li>
           <li><em>Can we understand the details of how water interacts with proteins — and use that to increase ligand potency?</em></li>
           <li><em>How do molecular structure, dynamics, and function change with cellular context?</em></li>
         </ul>
 
-        <p class="reveal">The answers are hidden in the data we collect today and will be
+        <p class="reveal">The answers are already hidden in the data we collect today and will be
           the central pursuit of the experiments of tomorrow. </p>
             
         <p class="reveal">To extract them, we need to develop
-          models that go beyond the static coordinates that we use to represent structure today. 
+          models that go beyond the static coordinates that we use to represent structure. 
           These models are hand-fit to experimental data, built by humans, for humans. They have been 
           incredibly successful: the Protein Data Bank (PDB) contains &gt;250k structures that produced
           9 Nobel prizes, guided &gt;80% of approved drugs, and trained AlphaFold, hailed as the first
           scientific breakthrough powered by AI. But lists of coordinates also limit the ability to
           reason about essential but complex aspects of structure: allostery, dynamics, water networks, 
-          partially ordered and disordered regions — in short, heterogeneity. When structure 
-          is compressed into a single coordinate model, this information is lost.
+          partially ordered and disordered regions — in short, heterogeneity.
           </p>
 
         <p class="reveal">We see a future in which AI models enable us to recapture this information,
           tackle the complexity of heterogeneity, and thereby fundamentally change how structure 
           impacts biology and medicine. Our vision is to ultimately bypass the coordinate models we
-          use currently and instead train AI models directly on structural experiments. In doing so, we
-          can leverage machine intelligence to do science beyond the limits of humans and coordinates.
+          use currently and instead interface AI models directly with structural experiments. In doing so, we
+          can leverage machine intelligence to do science beyond the limits of humans and the models we naturally understand.
         </p>
 
-        <p class="reveal">To see this future more concretely, take a practical problem that
-          faces drug hunters today: developing a molecular glue that creates a new
-          protein-protein interface between two proteins. This 
+        <p class="reveal">To see this future more concretely, consider one of the most challenging 
+          and exciting tasks facing drug hunters today: designing a molecular glue degrader that 
+          creates a new protein-protein interface between two proteins. This 
           task is essentially intractable for current rational, structure-based drug design. However, 
           with the assistance of a biomolecular foundation model capable of abstractly reasoning 
           about how small molecule binding induces structural changes, what drives protein-protein 

From b38665bddc4805e84be50cf771f17e6b9259734d Mon Sep 17 00:00:00 2001
From: tjlane <thomas.joseph.lane@gmail.com>
Date: Sat, 18 Jul 2026 10:32:45 -0400
Subject: [PATCH 20/22] @alisiafadini feedback pt 1

---
 ...sf.html => 2026-07-20-rss-joins-omsf.html} | 100 ++++++++----------
 .../rss-wordmark.png                          | Bin
 2 files changed, 44 insertions(+), 56 deletions(-)
 rename _posts/{2026-06-25-rss-joins-omsf.html => 2026-07-20-rss-joins-omsf.html} (84%)
 rename assets/posts/{2026-06-25-rss-joins-omsf => 2026-07-20-rss-joins-omsf}/rss-wordmark.png (100%)

diff --git a/_posts/2026-06-25-rss-joins-omsf.html b/_posts/2026-07-20-rss-joins-omsf.html
similarity index 84%
rename from _posts/2026-06-25-rss-joins-omsf.html
rename to _posts/2026-07-20-rss-joins-omsf.html
index 3c52d05..0f5062c 100644
--- a/_posts/2026-06-25-rss-joins-omsf.html
+++ b/_posts/2026-07-20-rss-joins-omsf.html
@@ -179,7 +179,7 @@
     <canvas class="fp-canvas" id="fp-diffraction"></canvas>
 
     <div class="fp-hairline"></div>
-    <div class="fp-date">June 25th 2026</div>
+    <div class="fp-date">July 20th 2026</div>
 
     <div class="fp-textstack">
       <div class="fp-textstack-inner">
@@ -190,7 +190,7 @@ <h1 class="fp-title">Structural Biology Has More to Teach AI Than Atomic Coordin
     </div>
 
     <div class="fp-logo">
-      <img src="{{ '/assets/posts/2026-06-25-rss-joins-omsf/rss-wordmark.png' | relative_url }}"
+      <img src="{{ '/assets/posts/2026-07-20-rss-joins-omsf/rss-wordmark.png' | relative_url }}"
            alt="Reciprocal Space Station" />
     </div>
   </div>
@@ -233,14 +233,18 @@ <h1 class="fp-title">Structural Biology Has More to Teach AI Than Atomic Coordin
         <p class="reveal">The data behind the structure — every diffraction pattern, scattering event,
           particle image — contains more information than can be captured in a single conformation.
           These data contain information about what molecules actually <em>do</em>: how they move,
-          fluctuate, bind, and respond to perturbation. To deposit coordinates, however,
-          experimentalists need to come up with the best single snapshot that explains the
-          data, treating real variability as noise. Tellingly, current structure
-          predictors struggle with experimental heterogeneity in part because they learned from this compressed
-          information. They trained on the world as it was deposited, not as it was measured.</p>
-
-        <p class="reveal">That lost information holds the key to some of the most important open
-          questions in molecular biology:</p>
+          fluctuate, bind, and respond to perturbation. Coordinate models, however, need to enumarate
+          all of that detail in an ever growing list of atom positions, which is possible for two or 
+          there of discrete conformations, but impossible for more continous motions and disorder.
+          Further, humans, who have to build these models by hand, have a limited ability to reason 
+          about  this complexity, and therefore seek the simplest, often single, conformation that 
+          explains as much of the data as possible, treating any variability — including functional
+          variation — as noise. Tellingly, current structure predictors struggle with experimental 
+          heterogeneity in part because they learned from this compressed information. They trained 
+          on the world as it was deposited, not as it was measured.</p>
+
+        <p class="reveal">This simplification causes us to lose information that holds the key to some
+          of the most important open questions in molecular biology:</p>
 
         <ul class="reveal">
           <li><em>When is a mutation harmless, and when does it reshape a functional state?</em></li>
@@ -250,8 +254,9 @@ <h1 class="fp-title">Structural Biology Has More to Teach AI Than Atomic Coordin
           <li><em>How do molecular structure, dynamics, and function change with cellular context?</em></li>
         </ul>
 
-        <p class="reveal">The answers are already hidden in the data we collect today and will be
-          the central pursuit of the experiments of tomorrow. </p>
+        <p class="reveal">Some of the answers to these questions are hidden in the data we already
+          have and collect today. Others will be the central pursuit of the experiments we invest in
+          tomorrow.</p>
             
         <p class="reveal">To extract them, we need to develop
           models that go beyond the static coordinates that we use to represent structure. 
@@ -265,9 +270,11 @@ <h1 class="fp-title">Structural Biology Has More to Teach AI Than Atomic Coordin
 
         <p class="reveal">We see a future in which AI models enable us to recapture this information,
           tackle the complexity of heterogeneity, and thereby fundamentally change how structure 
-          impacts biology and medicine. Our vision is to ultimately bypass the coordinate models we
-          use currently and instead interface AI models directly with structural experiments. In doing so, we
-          can leverage machine intelligence to do science beyond the limits of humans and the models we naturally understand.
+          impacts biology and medicine. Our vision is to eventually let AI models interface directly
+          with structural experiments, taking on the scale and complexity that current coordinate
+          models can't handle. In doing so, we can extend what's scientifically possible — uncovering
+          patterns and connections beyond what manual modeling allows — while freeing researchers to
+          focus their reasoning on the questions that matter most.
         </p>
 
         <p class="reveal">To see this future more concretely, consider one of the most challenging 
@@ -280,7 +287,7 @@ <h1 class="fp-title">Structural Biology Has More to Teach AI Than Atomic Coordin
           become routine.
         </p>
           
-        <p class="reveal">Surpassing the limits of static coordinates to realize this vision will require 
+        <p class="reveal">Surpassing the limits of coordinates to realize this vision will require
           a return to the measurements themselves. We need to build the AI-structure interface,
           the computational infrastructure that will make
           experimental data usable for inference of structural heterogeneity and enable training of
@@ -305,7 +312,7 @@ <h2 class="reveal">RSS &times; OMSF</h2>
           cryo-EM, diffuse scattering, and small-angle methods meet. We are setting out to explore
           this data space from our home base, the Reciprocal Space Station.</p>
 
-        <p class="reveal">RSS is part of the larger structural biology community, and is committed to
+        <p class="reveal">RSS is part of the larger structural biology community, and we are committed to
           supporting and sustaining it and ongoing structural software development. Stay
           tuned, as more details about how RSS plans to operate and contribute will follow in a
           future post!</p>
@@ -333,32 +340,17 @@ <h2 class="reveal">Building a path from measurements to function</h2>
           <a href="https://github.com/rs-station/embedopt">EmbedOpt</a> for observable-guided structure
           and ensemble inference.</p>
 
-        <p class="reveal">Together, these efforts define the RSS roadmap: make structural biology
-          data accessible and interoperable, implement the underlying physics as differentiable
-          models, and train the next generation of biomolecular AI on data, not structures.</p>
-
-        <p class="reveal">Concretely, RSS is building three connected layers.</p>
-
-        <p class="reveal">First, data libraries that make diffraction, scattering, imaging, and
-          time-resolved data easier to represent, manipulate, validate, and share. These are
-          building blocks that experimentalists, software developers, and model builders can use and
-          reuse.</p>
-
-        <p class="reveal">Second, experimental forward models that connect molecular structures and
-          ensembles to the measurements instruments actually record. These models need to be
-          physically correct, noise-aware, differentiable, performant, and usable across modalities
-          including crystallography, cryo-EM, SAXS, diffuse scattering, and beyond.</p>
-
-        <p class="reveal">Third, modeling systems that use experimental data directly: not only to
-          validate a final structure, but to guide structure determination, ensemble fitting, and
-          model training.</p>
+        <p class="reveal"><strong>Together, these efforts define the RSS roadmap: make structural
+          biology data accessible and interoperable, implement the underlying physics as
+          differentiable models, and train the next generation of biomolecular AI on data, not
+          structures.</strong></p>
 
         <h2 class="reveal">Where we are going</h2>
 
         <p class="reveal">We've already started down this path, with some exciting progress.</p>
 
         <h3 class="reveal">Ensemble fitting from raw observables</h3>
-        <p class="reveal">In dynamic regions, building a structural model by hand is often impossible. To address this challenge, we are developing engines that automatically fit distributions of conformations directly to experimental data. This will enable us to capture allostery, where changes on one side of a protein propagate to another, spatially removed site. Allostery opens new opportunities in drug discovery and is impossible to infer from a single-structure model. We are therefore extending our machinery to ensembles — stay tuned for an upcoming post!</p>
+        <p class="reveal">In dynamic regions, building a structural model by hand is often impossible. To address this challenge, we are developing engines that automatically fit distributions of conformations directly to experimental data. This will enable us to capture allostery, where changes on one side of a protein propagate to another, spatially removed site. Allostery opens new opportunities in drug discovery and is hard to capture with a single structure alone. We are therefore extending our machinery to ensembles — stay tuned for an upcoming post!</p>
 
         <h3 class="reveal">Data velocity for modern instruments</h3>
         <p class="reveal">Once interpretation no longer requires hours of manual model building per
@@ -372,31 +364,27 @@ <h3 class="reveal">From guiding existing models to training new ones</h3>
         <p class="reveal">Alongside sister projects such as
           <a href="https://openfold.io/">OpenFold</a>, we aim to close the
           experiment-model loop: measurements guide models, while models help design and interpret
-          experiments. We plan to build the infrastructure that will enable the next big
-          breakthroughs in biomolecular AI: cofolding, protein-protein interface prediction, and the
-          design of protein dynamics.</p>
+          experiments. We plan to build predictors that use experimental data directly, not only to
+          validate a final structure, but to guide structure determination, ensemble fitting, and
+          model training. We believe these will enable the next big breakthroughs in biomolecular AI:
+          cofolding, protein-protein interface prediction, and the design of protein dynamics.</p>
 
         <h2 class="reveal">Why OMSF</h2>
 
-        <p class="reveal">Infrastructure requires planning, maintenance, and resources. We'll need
-          to collaborate across structural biology, computation, simulation, drug discovery, and
-          experimental facilities. To succeed, our community will need to produce and support code
-          that is not only scientifically impactful, but available, reliable, extensible, and
-          community-driven.</p>
+        <p class="reveal">To realize our mission, scientific excellence is necessary but not
+          sufficient. Automation, performance, usability, documentation, testing, education, and
+          long-term stewardship are first-order concerns of RSS, alongside scientific insight. 
+          It will take collaboration across structural biology,
+          computation, simulation, drug discovery, and experimental facilities, and code that's not
+          just impactful, but available, reliable, extensible, and community-driven.</p>
 
-        <p class="reveal">That is why RSS is joining the Open Molecular Software Foundation to
-          further this mission in service of the structural biology and AI communities. OMSF
-          provides a home for open molecular software projects that support entire fields. Alongside
-          consortia such as <a href="https://openfold.io/">OpenFold</a>,
+        <p class="reveal">That's why RSS is joining the Open Molecular Software Foundation: a home
+          for open molecular software projects that support entire fields. Alongside consortia like
+          <a href="https://openfold.io/">OpenFold</a>,
           <a href="https://openforcefield.org/">OpenForceField</a>,
           <a href="https://openadmet.org/">OpenADMET</a>, and
-          <a href="https://openbind.uk">OpenBind</a>, RSS will help build
-          the open infrastructure needed for the next generation of structural biology.</p>
-
-        <p class="reveal">To realize this mission, scientific excellence
-          is necessary but not sufficient. Automation, performance, usability, documentation,
-          testing, education, and long-term stewardship are first-order concerns of RSS, alongside
-          scientific insight.</p>
+          <a href="https://openbind.uk">OpenBind</a>, RSS will help build the open infrastructure the
+          next generation of structural biology needs.</p>
 
         <h2 class="reveal">Join us</h2>
 
diff --git a/assets/posts/2026-06-25-rss-joins-omsf/rss-wordmark.png b/assets/posts/2026-07-20-rss-joins-omsf/rss-wordmark.png
similarity index 100%
rename from assets/posts/2026-06-25-rss-joins-omsf/rss-wordmark.png
rename to assets/posts/2026-07-20-rss-joins-omsf/rss-wordmark.png

From ca5fbcede4da22f6d3c19dee83f78b2aca5a4c52 Mon Sep 17 00:00:00 2001
From: tjlane <thomas.joseph.lane@gmail.com>
Date: Sat, 18 Jul 2026 10:42:57 -0400
Subject: [PATCH 21/22] @alisiafadini feedback pt 2

---
 _posts/2026-07-20-rss-joins-omsf.html | 15 ++++++++-------
 1 file changed, 8 insertions(+), 7 deletions(-)

diff --git a/_posts/2026-07-20-rss-joins-omsf.html b/_posts/2026-07-20-rss-joins-omsf.html
index 0f5062c..3a98a70 100644
--- a/_posts/2026-07-20-rss-joins-omsf.html
+++ b/_posts/2026-07-20-rss-joins-omsf.html
@@ -233,11 +233,11 @@ <h1 class="fp-title">Structural Biology Has More to Teach AI Than Atomic Coordin
         <p class="reveal">The data behind the structure — every diffraction pattern, scattering event,
           particle image — contains more information than can be captured in a single conformation.
           These data contain information about what molecules actually <em>do</em>: how they move,
-          fluctuate, bind, and respond to perturbation. Coordinate models, however, need to enumarate
-          all of that detail in an ever growing list of atom positions, which is possible for two or 
-          there of discrete conformations, but impossible for more continous motions and disorder.
+          fluctuate, bind, and respond to perturbation. Coordinate models, however, need to enumerate
+          all of that detail in an ever growing list of atom positions, which is possible for two or
+          three discrete conformations, but impossible for more continuous motions and disorder.
           Further, humans, who have to build these models by hand, have a limited ability to reason 
-          about  this complexity, and therefore seek the simplest, often single, conformation that 
+          about this complexity, and therefore seek the simplest, often single, conformation that
           explains as much of the data as possible, treating any variability — including functional
           variation — as noise. Tellingly, current structure predictors struggle with experimental 
           heterogeneity in part because they learned from this compressed information. They trained 
@@ -259,7 +259,7 @@ <h1 class="fp-title">Structural Biology Has More to Teach AI Than Atomic Coordin
           tomorrow.</p>
             
         <p class="reveal">To extract them, we need to develop
-          models that go beyond the static coordinates that we use to represent structure. 
+          models that go beyond the coordinates that we use to represent structure. 
           These models are hand-fit to experimental data, built by humans, for humans. They have been 
           incredibly successful: the Protein Data Bank (PDB) contains &gt;250k structures that produced
           9 Nobel prizes, guided &gt;80% of approved drugs, and trained AlphaFold, hailed as the first
@@ -340,7 +340,7 @@ <h2 class="reveal">Building a path from measurements to function</h2>
           <a href="https://github.com/rs-station/embedopt">EmbedOpt</a> for observable-guided structure
           and ensemble inference.</p>
 
-        <p class="reveal"><strong>Together, these efforts define the RSS roadmap: make structural
+        <p class="reveal"><strong style="font-weight: 600;">Together, these efforts define the RSS roadmap: make structural
           biology data accessible and interoperable, implement the underlying physics as
           differentiable models, and train the next generation of biomolecular AI on data, not
           structures.</strong></p>
@@ -390,7 +390,8 @@ <h2 class="reveal">Join us</h2>
 
         <p class="reveal">At its heart, RSS is an open community of reciprocal astronauts:
           structural biologists, software developers, experimentalists, method builders, and
-          partners who believe experiments have more to teach us than a single static model.</p>
+          partners who believe experiments have more to teach us than a single, static coordinate
+          model.</p>
 
         <p class="reveal">If you are excited about open software, experimental data, molecular
           dynamics, and the future of structural biology, reach out at

From 3ddc124d4cbd738c6b22bbcd89874058a7f085ad Mon Sep 17 00:00:00 2001
From: tjlane <thomas.joseph.lane@gmail.com>
Date: Sat, 18 Jul 2026 11:08:49 -0400
Subject: [PATCH 22/22] last touch ups

---
 _posts/2026-07-20-rss-joins-omsf.html | 19 +++++++++++--------
 1 file changed, 11 insertions(+), 8 deletions(-)

diff --git a/_posts/2026-07-20-rss-joins-omsf.html b/_posts/2026-07-20-rss-joins-omsf.html
index 3a98a70..da59dd3 100644
--- a/_posts/2026-07-20-rss-joins-omsf.html
+++ b/_posts/2026-07-20-rss-joins-omsf.html
@@ -234,11 +234,15 @@ <h1 class="fp-title">Structural Biology Has More to Teach AI Than Atomic Coordin
           particle image — contains more information than can be captured in a single conformation.
           These data contain information about what molecules actually <em>do</em>: how they move,
           fluctuate, bind, and respond to perturbation. Coordinate models, however, need to enumerate
-          all of that detail in an ever growing list of atom positions, which is possible for two or
-          three discrete conformations, but impossible for more continuous motions and disorder.
-          Further, humans, who have to build these models by hand, have a limited ability to reason 
-          about this complexity, and therefore seek the simplest, often single, conformation that
-          explains as much of the data as possible, treating any variability — including functional
+          all of that detail in an ever-growing list of atom positions, which is possible but already
+          challenging for two or three discrete conformations. At the frontier of structural science
+          are continuous motion and disorder — here, coordinate models become intractable: the number
+          of parameters explodes beyond the power of the data to determine them, and the resulting
+          models inhibit rather than aid scientific interpretation.</p> 
+          
+        <p class="reveal">As a consequence, structural biologists seek the simplest,
+          often single, conformation that explains as much of the data as possible, treating any 
+          variability — including functional
           variation — as noise. Tellingly, current structure predictors struggle with experimental 
           heterogeneity in part because they learned from this compressed information. They trained 
           on the world as it was deposited, not as it was measured.</p>
@@ -258,9 +262,8 @@ <h1 class="fp-title">Structural Biology Has More to Teach AI Than Atomic Coordin
           have and collect today. Others will be the central pursuit of the experiments we invest in
           tomorrow.</p>
             
-        <p class="reveal">To extract them, we need to develop
-          models that go beyond the coordinates that we use to represent structure. 
-          These models are hand-fit to experimental data, built by humans, for humans. They have been 
+        <p class="reveal">To extract them, we need to go beyond the coordinates that we use to represent structure. 
+          Coordinate models are hand-fit to experimental data, built by humans, for humans. They have been 
           incredibly successful: the Protein Data Bank (PDB) contains &gt;250k structures that produced
           9 Nobel prizes, guided &gt;80% of approved drugs, and trained AlphaFold, hailed as the first
           scientific breakthrough powered by AI. But lists of coordinates also limit the ability to