griffithlab / docm Goto Github PK

You example includes a drug_interactions field in the response, but if you try this example no drug interaction data is returned.

/api/v1/variants/ENST00000263967:c.1633G>C.json

If I go to http://docm.genome.wustl.edu/about, it redirects me to http://www.docm.infoabout/ (which is not a thing)

Missing slash in the redirect rules somewhere, I'm guessing

A user found the following discrepancy

There is a set of discrepancies between the VCF and TSV files in the DOCM.
In the TSV, there are four 4bp insertions in the NPM1 gene (chr5, 170837546).
In the VCF, they are listed as 3bp insertions.
Based on the cited paper (PMID 19657110), I believe that the TSV version is correct.

This column is not actually showing pubmed IDs, we should name it more descriptively. "Citations" is my vote but open to suggestions.

We should license DoCM the same way that we license CIViC. Creative Commons Attribution 4.0 International License.

If a variant isn't found the API returns 404, it would be useful to return something that indicates your request was good but the variant wasn't found.

The HGVSc. annotations in the TSV file are all given with respect to the plus strand, but they should be given with respect to the sense strand. For example, the variant in MTOR at chr1, 11174420 is listed as c.7255C>T, but it should be c.7255G>A (since the coding strand of MTOR is the minus strand).

Linking in from other resources to DoCM for gene-level information is currently unavailable. It would be nice if I could go to docm.genome.wustl.edu/genes/PTEN and see the webpage rendered like so:

It would be great if the was a link to the CIViC variant summary in the "Variant Data" box of the variant page.

Additionally, a link to the accompanying evidence item could go into the external links column of the disease data box on the variant page.

It would be especially cool if this was a icon or button with the civic logo. Maybe "see in CIViC" or something.

How is a list of pubmed IDs currently sorted in DoCM. e.g. PIK3CA p.E545K lists these pubmed ids:

18725974, 22162589, 19513541, 19029981, 16906227, 21430269, 15647370, 22271473, 22162582, 15805248, 20453058, 18676830, 15254419, 15016963, 16930767, 20619739, 19903786, 19366826

Maybe these should be ordered from oldest to newest?

Also, this list of IDs is very important to have, but not very informative to view. Would it be better to list the publications in the format of Jones et al. 2015. So that at least for some people they would be more familiar.

e.g. (filtered from 1,015 total entries)

Display the number of variants from the current version.

TravisCI fails when trying to install better_errors referencing a requirement of Ruby version >= 2.0.0. @acoffman could that be fixed by changing the following line to 2.0.0?

The following variants are not importing

chromosome  start   stop    gene    reference   variant
4   55593594    55593594    KIT GAAGTACAGTGGAAG -
4   55593594    55593594    KIT GAAGTACAGTGGAAGGTT  -
4   55593600    55593601    KIT -   CTA
4   55593600    55593601    KIT -   CTC
4   55593600    55593601    KIT -   CTG
4   55593600    55593601    KIT -   CTT
4   55593600    55593601    KIT -   TTA
4   55593600    55593601    KIT -   TTG
4   55593609    55593611    KIT GTT -
4   55593628    55593628    KIT G   T
4   55593630    55593647    KIT AACAATTATGTTACAGAC  -
4   55593630    55593656    KIT AACAATTATGTTACAGACCCAACA    -
4   55593657    55593658    KIT -   CCAGAA
4   55593657    55593658    KIT -   CCAGAG
4   55593657    55593658    KIT -   CCCGAA
4   55593657    55593658    KIT -   CCCGAG
4   55593657    55593658    KIT -   CCGGAA
4   55593657    55593658    KIT -   CCGGAG
4   55593657    55593658    KIT -   CCTGAA
4   55593657    55593658    KIT -   CCTGAG
4   55593669    55593671    KIT GAT -
4   55594197    55594197    KIT C   T
4   55599338    55599338    KIT A   T
4   55602700    55602700    KIT A   G
4   55602770    55602770    KIT C   T
7   6426892 6426892 RAC1    C   T
7   128829195   128829196   SMO CC  TT
7   128845101   128845101   SMO C   T
7   128846116   128846117   SMO CC  TT
7   128849224   128849225   SMO CC  TT
7   128850279   128850280   SMO CC  TT
7   128850341   128850341   SMO G   T
7   128850838   128850838   SMO G   A
7   128851883   128851883   SMO C   T
7   128852191   128852192   SMO CC  TT
7   140453155   140453155   BRAF    C   A
9   133738307   133738307   ABL1    A   T
19  17948006    17948006    JAK3    G   A
19  17948009    17948009    JAK3    G   A
19  18271909    18271909    PIK3R2  C   T

Please and thank you!

Do any search (I pulled down 'acute myeloid leukemia'), then export at TSV of the list. The first line of the text file contains the following:
ENST00000373103:c.1853G>A

Instead of the expected
ENST00000373103:c.1853G>A

I find the lack of spacing between news items a bit jarring:

http://docm.genome.wustl.edu/news

Would be nice if the syle looked a bit more like this:
http://dgidb.genome.wustl.edu/news

Right now CIViC and DGIdb have one, but DoCM does not.

Add VEP annotation and account for tags and rationale

The current about page http://docm.genome.wustl.edu/about is very light on content.

This should be a much richer description of the goals of DOCM. Perhaps include a visual as we do in CIViC: https://civic.genome.wustl.edu/#/collaborate

Inspiration for the content of the about page could come from the manuscript.

In the filter options at the right side of the page it would be awesome if we could ask for all variants associated with a particular publication (e.g. pubmed ID).

Currently, all transcripts in DoCM are labeled as build 74 of ensembl genes. Is this reflective of the VEP annotations we're pulling? My understanding is that the API only hits the most recent build... e74.rest.ensembl.org isn't a thing, unfortunately. I suggest that we check that the coordinates of our existing variants in DoCM are correct in 84, and then update our tags accordingly.

Currently when you look at a variant in DOCM, there is a link out to results in DGIdb like this:

Gene: KRAS ( View drug interactions on DGIdb )

We should add a similar link to the gene summary page in CIViC

The current sources page (http://docm.genome.wustl.edu/sources) is worded in such a way that it appears that the only/main sources for DoCM are MyCancerGenome and the Knowledge Database. This is an underrepresentation of the sources that inspired content for DoCM. We should better explain how DoCM uses such sources, as well as key papers that published extensive lists, and more focused papers that describe specific groups of variants.

The news page has not been kept very up to date with developments. We should summarize when new features are added, when new variants are added, and perhaps even when important bugs are fixed. We can also report cases where DoCM will be presented publicly here.

The current FAQ page feels a bit like a place holder right now. We should fill this page in with more content. The FAQ pages of DGIdb and CIViC may provide some inspiration.

@jmcmichael, When you have a minute could you change the TGI logo to MGI?

Thanks!

variants like http://docm.genome.wustl.edu/variants/ENST00000261609:c.2277delAACGGTCCTGACCTG?version=2

screw up the formatting.

@acoffman could this be incorporated in the next release?

It would be really nice to have a list of all the versions of the database, perhaps on the about page?

It could be a table with the version number the number of mutations in that version, the number of publications, and the number of cancer subtypes (DOIDs)

We should move the 'Chromosomes' filter down so that it is directly above the start/stop 'Position' filter. Then only allow filtering on position if one or more chromosomes has been selected?

Some additional variant lists have been curated but have not yet been imported into the liver version of DoCM. Prior to publication we should do these imports and update the News feed as we add them.

a variant can have many tags.

Tags must be 5 words or fewer.

Tags can describe the variant and allow for custom organization and filtering

We don't want our users to leave DoCM and never come back! It would be nice if a new tab would open when you click on a link like the pubmed id, dgidb or mycancergenome.

Users should submit variants in a TSV format

Add a field that allows the submitter to describe their rationale for including the batch in DoCM. This should be required and viewable on the interface following acceptance.

We should create a DoCM tutorial. This could live on a new page of the DoCM site (e.g. help or getting started).

This tutorial could include things like:

• Query using DoCM website to get all variants for a specific disease, all variants for a specific gene, all variants for a region of the genome.
• Export results of a query to a VCF file or TSV file
• Use the API to perform the same queries. Provide an example script that does this. To lower the bar for a broad variety of users, provide examples written in Perl, Python and Ruby.

This could exist under a "pending" version, just like "current" version.

There should be some indication if a variant disease or publication is pending review.

This would allow for more routine updates of CIViC data without re-importing a version from scratch.

Hello, the vcf download has "/" and " " in the info field and this breaks at least pyvcf, does this fit VCF spec?
Thanks

Matt

The current (most recent) version is at the top, and then the versions are listed in ascending order. It seems more intuitive to me that the versions would be listed in descending order, with the second entry starting at the penultimate version.