Reference Database used to annotate genes

[anjalijain22]

Hello, hope you're doing well! I've been encountering some challenges while creating gene lists in our seqr instance. When I paste a gene list into seqr, I receive numerous "invalid gene and interval" error messages. Upon investigation, I noticed that many genes in my original query have another gene symbol in seqr. For example, the gene SKIC3 appears as TTC37 in Seqr.

Additionally, some genes like RNU12 are accepted when creating the list, but when I attempt to use the list on the variant search page (specifically in the GRCh37 dataset), I get an error:
"Invalid gene interval: RNU12"
I understand this is likely due to RNU12 failing the hg19 liftover, but this isn’t immediately obvious unless I manually search for the gene in the gene info section and check its coordinates.

I had a couple of questions that might help streamline this process on our end:

• What is the source database seqr uses to annotate gene names? Is it downloadable from seqr, or is there a recommended method to convert gene names to ones that seqr recognizes?
• How frequently is this database updated?

Any guidance or suggestions would be greatly appreciated!

[hanars]

seqr uses gencode as the source for genes. seqr is not designed to be a reference data source and we do not recommend "downloading" reference data from seqr - if you wish to interrogate what is in the seqr database the better way to do that would be to query the database directly.

Assuming you are using the helm chart we provide without any customization, reference data should automatically update whenever new reference data is available, and it should be checking for new data every 5 minutes: https://github.com/broadinstitute/seqr-helm/blob/main/charts/seqr/values.yaml#L103-L105
However, we pin the versions of gencode that seqr uses so it will not update unless there is an explicit code change to do so. This is because we want to ensure consistency between the gencode version used when VEP is run in the loading pipeline and variants are annotated in genes and the genes we show in seqr. There is no real mechanism for updating all of VEP, so updating gencode is risky. Currently, the latest included version is 39: https://github.com/broadinstitute/seqr/blob/master/reference_data/models.py#L242-L243

[anjalijain22]

Hi Hana, thanks for this information! I have another quick follow up question - I noticed that some genes have two entries in the seqr reference database, eg. RPS4Y2. The gene ID ENSG00000280969 corresponds to Gencode 39 and ENSG00000157828 corresponds to Gencode 19. The GRCh37 coordinates for the gene differ slightly for the two gene IDs. If I query this gene in a GRCh37 WES dataset, does the variant search engine default to using the coordinates of the gene ID that corresponds to Gencode 19 or to Gencode 39?

[hanars]

When performing a gene search seqr will default to using the latest available gencode release for the given gene symbol

[anjalijain22]

Got it! Thank you so much Hana!