How to contribute ontology subtrees?

[sjett-biohub]

Hello,

We are interested in learning more about how to submit not just individual terms to FBbi, but potentially subtrees. We have two areas where we are building terminology trees, cryoelectron tomography and optical pooled screening.

In the case of cryoET, our cryoET team, in collaboration with others, have developed an ontology that describes cryoET experiments and the conditions for those experiments. Most of the terms have broad application in cryoET experiments, and some are currently specific to experiments performed at the Chan Zuckerberg Imaging Institute. Data submitted to the CryoET Data Portal is harmonized to this ontology in preparation for use of Portal data for AI training. If you are interested, I can share their ontology.

Others in our organization are working on an ontology for optical pooled screening (OPS), a method for screening genetic manipulations through high throughput microscopy. We have already run into a problem OPS mentioned in issue #36 for CLEM, multiple terminologies for a single method. We are still trying to devise a solution for our situation, but it is heartening to see it being discussed for CLEM.

I’ve been reading the recent discussion about revising the subClassOf hierarchy (issue #50), and am glad to see this being discussed, along with the discussion regarding the use of terms from other ontologies (issue #39). We see both issues as important as we try to work with others to construct ontologies that meet both the needs of the broader community and specialized needs specific to our research. For example, we noticed the cross-referencing of several terms from the CHMO ontology, but also noticed additional terms in CHMO that have relevance to FBbi (for example, transmission electron microscopy grid). In the cryoET ontology mentioned above, TEM grids play a major role in cryoET, and we have a sub hierarchy of terms related to the grids we would be happy to contribute.

The mention in #39 of probes is one where there seems to be a wide variety of opinions amongst researchers. Is the probe a reporter ligand that highlights the presence of a target, or is the probe the molecule that binds to a target, and is bound to a reporter ligand? It appears to be the former in the discussion, but that potentially leads to an additional subclass for binding molecules.

My apologies for the long post. Summary, we are interested in contributing a series of terms for trees related to methods important to a number of researchers, and are asking about the best next steps.

Thanks to all of you for your work on maintaining the FBbi.

Best regards,
Steve Jett
Biohub

P.S. If anyone will be at the FoundingGIDE meeting, my colleagues Jason Swedlow and Chi-Li Chiu will be there in person. Chi-Li is also working on metadata, ontologies and is spearheading our efforts on OME-Zarr. I will virtually attend the meeting.

[gouttegd]

Thank you for your interest!

we are interested in contributing a series of terms for trees related to methods important to a number of researchers, and are asking about the best next steps.

We’re definitely interested in accepting new terms that people feel are currently missing in FBbi.

If you already have a precise list of terms you’d want to see in FBbi, the next best step would be to make that list available to us, in whatever form is most convenient for you.

No problem a priori about integrating hierarchies of new terms, instead of individual terms.

I will unfortunately miss the FoundingGIDE meeting, though I may try to join remotely.

[sjett-biohub]

Hello, that's wonderful news!

For cryo-ET terminologies, my colleagues have devised a term tree that meets most of their needs. The tree, which may also be found at CDPO - CryoET Data Portal Ontology, is:

BFO:entity
└── BFO:continuant
└── BFO:independent continuant
├── BFO:material entity
│ ├── OBI:specimen
│ ├── CDPO:sample
│ ├── CDPO:sample support component
│ │ ├── CDPO:sample support material
│ │ ├── CDPO:electron microscopy grid - (CHMO:continuant:independent continuant:material entity:processed material:device:transmission electron microscope grid)
│ │ └── CDPO:support film
│ │ ├── CDPO:carbon support film
│ │ │ ├── CDPO:holey carbon film
│ │ │ └── CDPO:lacey carbon film
│ │ ├── CDPO:gold support film
│ │ │ └── CDPO:holey gold film
│ │ └── CDPO:support film fragment
│ ├── CDPO:contamination
│ │ ├── CDPO:ice particle contamination
│ │ ├── CDPO:sputter particle contamination
│ │ └── CDPO:leopard skin contamination
│ ├── CDPO:sample artifact region
│ │ ├── CDPO:non-vitreous region
│ │ ├── CDPO:dose-damaged region
│ │ ├── CDPO:curtaining artifact region
│ │ └── CDPO:sample tear
│ ├── CDPO:fiducial marker
│ │ ├── CDPO:gold fiducial nanoparticle
│ │ └── CDPO:gold CLEM label
│ ├── CDPO:protein aggregate
│ └── BFO:fiat object part
│ └── CDPO:tomogram edge
└── BFO:immaterial entity
└── BFO:site
└── CDPO:vacuum region

IAO:information content entity
└── IAO:image
└── CDPO:tomographic image artifact
├── CDPO:tomographic reconstruction artifact
│ ├── CDPO:unreconstructed region
│ └── CDPO:streak artifact
├── CDPO:sample charging artifact
├── CDPO:laser phase plate off-plane artifact
└── CDPO:laser phase plate off-node artifact

I noted where a term already exists in CHMO, and bolded two terms at the bottom which currently would be specific imaging performed at our organization (Biohub), but will likely increase in importance and adoption in the future. There is also some potential overlap with the ongoing discussion on CLEM probes, and probes in general.

We realize this is a lot, represents a lot of work on the part of my colleagues and will probably require a multifaceted approach to contribute new terms, adopt terms from other ontologies, and stimulate discussions on how to approach terms with broad interest, but also wide definitions (probes again, for example). Our goal is to find out how we can best contribute. Thank you for your consideration.

[gouttegd]

Thank you for sharing your CDPO ontology!

Looking at it, I see that some branches would indeed be relevant in FBbi, notably sample support component and fiducial marker. (The other branches however seem out-of-scope.)

We realize this is a lot, represents a lot of work on the part of my colleagues and will probably require a multifaceted approach to contribute new terms, adopt terms from other ontologies, and stimulate discussions on how to approach terms with broad interest, but also wide definitions (probes again, for example). Our goal is to find out how we can best contribute.

“Stimulating discussions” is in itself of great value. :)

In this particular case, since the CDPO ontology already exists, my preferred approach would be to create in FBbi a handful of terms that can serve as “branching points” to which existing CDPO terms can be attached.

That is, we would create for example a single sample support component term somewhere in our hierarchy (where exactly remains to be determined, and that will probably have to wait until we are done reorganising our current hierarchy as mentioned in this comment), and we would mark that term as being equivalent to CDPO’s own sample support component term. This would be understood as meaning, “if you need a more precise term to refer to a specific type of sample support component, go look at this particular branch of CDPO”.

This would avoid a needless duplication of effort between FBbi and CDPO, while ensuring that people have all the terms they need.