UCSC-Fiber-seq-hub

A UCSC Track Hub for hg38 exposing FIRE (Fiber-seq Inferred Regulatory Elements) chromatin accessibility and peak calls across a compendium of cell lines and tissues.

Hub URL: https://fiberseq.github.io/UCSC-Fiber-seq-hub/hub.txt

Load it via My Data → Track Hubs → My Hubs, or append &hubUrl=<hub URL> to any hgTracks URL (add &udcTimeout=1 while iterating).

Open it directly on the UCSC test browser: genome-test.gi.ucsc.edu

Tracks

Both tracks sit under one collapsible Fiber-seq superTrack in the hub's own group — UCSC lists hub tracks in their own section below the browser image, separate from the native groups (they can't be merged in next to ENCODE's cCREs) — and each has its own description page.

• Fiber-seq Accessibility (on by default): percent-accessible chromatin for the 7 default cell lines, overlaid as semi-transparent bars in a colorblind-safe palette.
• Fiber-seq Compendium: a faceted composite giving every sample three subtracks — percent accessible (both haplotypes), FIRE peaks, and a Hap1/Hap2 overlay (haplotype 1 blue, haplotype 2 red). Hidden overall by default; an enabled subtrack renders dense. Filter and sort by Cell Type and Track Type; those facets and the per-track-type colors come from the hg38/fireCompendium_metadata.tsv and hg38/fireCompendium_colors.json sidecars the build emits.

Raw bigWig/bigBed files live on Kopah S3 as per-sample hubs; this repo only builds the merged hub.txt/genomes.txt/trackDb.txt pointing at them. No sequencing data is stored here.

Build

The tracking sheet is the master sample list. scripts/build_hub.py (stdlib-only) fetches it live, templates each sample's file URLs, HEAD-checks them, and drops any unreachable file rather than shipping a broken link. A row is included once it has a name, tissue, and working "total % accessible" link. Only the sheet's first published tab is read, so rows parked in an unused tab are ignored.

pixi install            # one-time
pixi run build          # regenerate hub/ from the sheet
pixi run validate       # rebuild + hubCheck -noTracks (run before pushing)
pixi run validate-full  # also re-fetches every bigWig/bigBed

Everything under hub/ is committed and is exactly what deploys to GitHub Pages: hub.txt, genomes.txt, hg38/trackDb.txt, the two description pages, the compendium's hg38/fireCompendium_metadata.tsv and hg38/fireCompendium_colors.json faceting sidecars, and samples.tsv — a machine-parseable manifest of what shipped (one row per sample; columns for name, PS ID, tissue, default flag, and each track type's data-file URL), served at .../samples.tsv. genomes.txt stamps the trackDb URL with a content hash (trackDb.txt?v=<hash>) so UCSC can't get stuck on a stale cached copy.

To add a sample: fill in its name, tissue, and links in the sheet, then pixi run validate and commit the hub/ changes.

CI runs the fast -noTracks check on every push and before deploy. Before submitting to UCSC, switch deploy.yml to the full -udcDir check (see the TODO there) so a broken data link can't go live.

Citation

Vollger, M. R.*†, Swanson, E. G.*, Neph, S. J., et al., Stergachis, A. B.† (2026). Somatic epimutations cap genetic determinism in the human diploid chromatin epigenome. Cell, in press. Preprint

Fiber-seq method: Stergachis, A. B., et al. (2020). Single-molecule regulatory architectures captured by chromatin fiber sequencing. Science 368(6498):1449–1454. https://doi.org/10.1126/science.aaz1646

Contact

Vollger Lab, University of Utah (mrvollger@genetics.utah.edu) · Stergachis Lab, University of Washington (absterga@uw.edu)