Variant normalization: e.g. Assembly identifier (GRC notation, e.g. `GRCh37`)

Question

Variant normalization: e.g. Assembly identifier (GRC notation, e.g. `GRCh37`)

bwalsh opened this issue 7 years ago · comments

I've been looking into adding a simple GA4GH beacon.
As part of that I discovered a new area for Harmonization - reference genome

From ga4gh beacon:
https://github.com/ga4gh/beacon-team/blob/develop/src/main/resources/avro/beacon.avdl#L76
What we have now

References:
https://genome.ucsc.edu/FAQ/FAQreleases.html#release1

Issue: What standard harmonization do we need to apply?

@jgoecks @ahwagner : thoughts?

Jeremy Goecks · Answer 1 · Wed Oct 18 2017 00:19:13 GMT+0800 (China Standard Time)

@bwalsh @ahwagner Two suggestions:

Standardize on GRC naming conventions, namely GRCh37 and GRCh38
For all variant coordinates (start, stop), provide them in GRCh37. Time/priority permitting, add coordinates for GRCh38.

This is actually a big deal because GENIE called variants against GRCh37, so GRCh38 variants won't match and/or will yield spurious matches.

Alex H. Wagner, PhD · Answer 2 · Wed Oct 18 2017 01:08:15 GMT+0800 (China Standard Time)

Yeah, I agree with @jgoecks re: GRC naming conventions. For queries to the web app / API, we may want to consider either:

infer the reference from submitted variants (hard)
enforce selection of a reference at query (easy, but requires additional work from users)

...I prefer the easy route, as we'll have other tough tasks ahead of us.

Malachi Griffith · Answer 3 · Wed Oct 18 2017 02:59:58 GMT+0800 (China Standard Time)

Should we consider using the Allele Registry to normalize incoming variants?

Then use that to represent variants various ways (build37 vs build38, genome level vs transcript level, refseq transcripts vs ensembl vs LRG transcripts, etc.)?

http://reg.clinicalgenome.org/

Here is an example variant allele returned by various approaches from the registry:
http://reg.clinicalgenome.org/redmine/projects/registry/genboree_registry/by_caid?caid=CA000301
http://reg.clinicalgenome.org/redmine/projects/registry/genboree_registry/allele?hgvs=NC_000017.10:g.7578212G%3EA
http://reg.clinicalgenome.org/redmine/projects/registry/genboree_registry/allele?hgvs=NC_000017.11:g.7674894G%3EA

Brian · Answer 4 · Sat Oct 21 2017 06:50:52 GMT+0800 (China Standard Time)

@jgoecks @ahwagner : refactored results

Note: cgi is missing, I'm following up with David T

Brian · Answer 5 · Sat Oct 21 2017 08:01:05 GMT+0800 (China Standard Time)

+cgi
switched brca to 37

Brian · Answer 6 · Sat Dec 02 2017 09:35:07 GMT+0800 (China Standard Time)

Here is the status for v0.7 (data freeze)

all sources are enriched with feature information based on a constructed hgvs genomic location g.
hit rates, per source are documented below

The enriched data is distilled into two fields feature.links and feature.synonyms.

For example for civic variant EML4-ALK C1156Y
the links and synonyms attributes are sourced from allele_registry.

Of course, there is more information that could be merged. Propose waiting for specific use cases.

  {
  "entrez_id": 238,
  "end": 29445258,
  "name": "EML4-ALK C1156Y",
  "links": [
    "http://myvariant.info/v1/variant/chr2:g.29222392C>T?assembly=hg38",
    "http://reg.genome.network/refseq/RS000050",
    "http://www.ncbi.nlm.nih.gov/clinvar/?term=363219[alleleid]",
    "http://reg.genome.network/refseq/RS000026",
    "http://reg.genome.network/allele/CA16602783",
    "http://reg.genome.network/refseq/RS000002",
    "http://www.ncbi.nlm.nih.gov/clinvar/variation/376340",
    "http://cancer.sanger.ac.uk/cosmic/mutation/overview?id=99136",
    "http://myvariant.info/v1/variant/chr2:g.29445258C>T?assembly=hg19",
    "http://reg.genome.network/refseq/RS001597"
  ],
  "start": 29445258,
  "synonyms": [
    "NC_000002.10:g.29298762C>T",
    "chr2:g.29445258C>T",
    "LRG_488:g.704175G>A",
    "NC_000002.12:g.29222392C>T",
    "NG_009445.1:g.704175G>A",
    "CM000664.2:g.29222392C>T",
    "chr2:g.29222392C>T",
    "NC_000002.11:g.29445258C>T",
    "CM000664.1:g.29445258C>T",
    "COSM99136"
  ],
  "biomarker_type": "snp",
  "referenceName": "GRCh37",
  "geneSymbol": "ALK",
  "alt": "T",
  "ref": "C",
  "chromosome": "2"
}

--//--

allele registry hit rates

source	Total	filters	Count
molecularmatch_trials	199069	all features w/ location	77765
molecularmatch_trials		allele_registry (hit)	76368
molecularmatch_trials		allele_registry (miss)	1397
jax	5754	all features w/ location	4894
jax		allele_registry (hit)	4893
jax		allele_registry (miss)	1
brca	5717	all features w/ location	5717
brca		allele_registry (hit)	5055
brca		allele_registry (miss)	662
oncokb	4048	all features w/ location	1796
oncokb		allele_registry (hit)	1792
oncokb		allele_registry (miss)	4
civic	3497	all features w/ location	3497
civic		allele_registry (hit)	1704
civic		allele_registry (miss)	1793
molecularmatch	2079	all features w/ location	1665
molecularmatch		allele_registry (hit)	1421
molecularmatch		allele_registry (miss)	244
cgi	1432	all features w/ location	1432
cgi		allele_registry (hit)	589
cgi		allele_registry (miss)	843
jax_trials	1173	all features w/ location	1094
jax_trials		allele_registry (hit)	1094
jax_trials		allele_registry (miss)	0
pmkb	609	all features w/ location	609
pmkb		allele_registry (hit)	0
pmkb		allele_registry (miss)	609
sage	69	all features w/ location	0
sage		allele_registry (hit)	0
sage		allele_registry (miss)	0