BEP-0006 - Gene Modifications

[cthoyt]

This BEP proposes an addition to the BEL 2.0+ syntax to support gene modifications (gmod())

The full text can be found at: https://github.com/belbio/bep/blob/bep-0006/docs/drafts/BEP-0006.md

[wshayes]

What about genomic modifications that are not directly attributable to a gene (e.g. a SNP)? Can we come up with a way to capture both genomic and gene modifications that are impactful?

[cthoyt]

Here's an example when we encoded a variant (with minimal information about actually what it is, that we hope we could look up from dbSNP later)

SET Citation = {"PubMed", "29183403"}

SET Evidence = "
    Rs142787485 in RAB10 confers significant protection 
    against AD (p value = 0.0184, odds ratio = 0.5853). Moreover, we replicated 
    this association in an independent series of unrelated individuals 
    (p value = 0.028, odds ratio = 0.69) and used a gene-based test to confirm a
    role for RAB10 variants in modifying AD risk (p value = 0.002). Experimentally,
    we demonstrated that knockdown of RAB10 resulted in a significant decrease
    in Aβ42 (p value = 0.0003) and in the Aβ42/Aβ40 ratio (p value = 0.0001) in 
    neuroblastoma cells. We also found that RAB10 expression is significantly 
    elevated in human AD brains (p value = 0.04).
"

g(HGNC:RAB10) hasVariant g(dbSNP:rs142787485)
g(dbSNP:rs142787485) neg path(MESHD:"Alzheimer Disease")

p(HGNC:RAB10) -> p(HGNC:APP, frag(672_713))
p(HGNC:RAB10) pos path(MESHD:"Alzheimer Disease")

Otherwise, I don't see why we can't refer to SNPs with the geneAbundance() with the language as is (assuming there's a namespace for dbSNP, perhaps with a regular expression)

[ncatlett]

1. Rationale and Goals - by gene acetylation, do you mean histone acetylation?
2. Use cases - can you clarify "effects on several levels"? It is not obvious (to me) what this means, and there is only one example.
3. Discussion - For histone proteins, can this be modeled as a modified histone protein/gene complex? It seems different than a gene modification
4. Discussion - Yes, we absolutely need to enable specification of position information. This should handle the type of position information reported in the literature (and resources like ENCODE), and also enable the mapping of data to the BEL knowledge networks (RRBS, Illumina methylation array, etc.). I'm guessing that ideally we'd want to capture this at a few levels - specific nucleotide (coordinates), specific DNA region (coordinates), named region, e.g., "promoter".
5. Specification - are there any implications to reusing the default BEL namespace protein modification terms "methylation" for gene modifications? Is there an alternative to using the GO Biological Process terms under "DNA Modification" (am worrying about keeping separate the process of modification vs. the modification itself).

[cebel]

From my understanding of Semantic Versioning the extension of BEL specification with a new function like gmod obligatorily needs an increase in the MINOR version number to 2.1.0 and not 2.0.0.+ (perhaps this was the meaning of the plus?). I suggest also, that BEL should try to follow strictly the NAMESPACE:NAME convention without another set of arguments for a new function.

[wshayes]

Hi Christian, I proposed using the 2.0.0+ in order to not determine specifically that a feature will go into a minor release or a major release. I agree that gmod() is a minor release feature - it does not mean that we will bundle it into something that is backwards compatible which would require a major release. Also, BEP’s are not handled in order so saying that it should be 2.1.0 is misleading as it may wind up in 2.3.0 for example. What we need to describe in that part of the BEP is: Is this backwards compatible? Does it target a specific version of BEL? 2.0.0+ indicates it targets versions beyond 2. Thanks for reviewing it - hope this clarifies our thinking a little bit more. Please feel free to add a Pull Request to the BEP Template to make this clearer to you and others. Wm

…

On Aug 6, 2018, at 2:27 AM, Christian Ebeling ***@***.***> wrote: From my understanding of Semantic Versioning <https://semver.org/> the extension of BEL specification with a new function like gmod obligatorily needs an increase in the MINOR version number to 2.1.0 and not 2.0.0.+ (perhaps this was the meaning of the plus?). I suggest also, that BEL should try to follow strictly the NAMESPACE:NAME convention without another set of arguments for a new function. — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#13 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAWS9_DuxgsiGr4p07L-wduAlE5s0YqBks5uN-HGgaJpZM4VSuBB>.

[johnbachman]

Relevant proposal from HGVS: http://varnomen.hgvs.org/bg-material/consultation/svd-wg005/

[cthoyt]

Discussion Notes

How much of a gene is modified, and in what position?

HGVS has an open issue for methylation (see https://varnomen.hgvs.org/recommendations/open-issues/). Seems like we'd have our own syntax for this one.

Idea: add a range as a second optional argument

Other thing to think about: what about the promoter?

X binds promoter of Y and X increases expression of Y -> X is a transcription factor of Y

John mentions: contact stuff?

What rules should there be for gene modifications? Things that are localized to a certain gene. If it's global, there's probably a GO process describing it.

Can we show that this gene region is equivalent to the gene's promoter?

TODO: different scenarios for locations, ranges, and specific parts like promoter.