ClinVar Annotation Updates for seqr local installations

[anjalijain22]

Hi,

Recently, our team noticed some discrepancies between the ClinVar annotations shown on seqr versus annotations on ClinVar. For example: This variant is annotated as "likely pathogenic" on seqr:

But on ClinVar, it is annotated as likely Benign: https://www.ncbi.nlm.nih.gov/clinvar/variation/986331/?oq=974463[alleleid]&m=NM_014585.6(SLC40A1):c.689C%3EA%20(p.Thr230Asn)

I suspect this is because the "likely benign" submission to ClinVar was done relatively recently (Oct 19, 2022) and we are using an older version of the ClinVar Hail matrix table from where seqr fetches its ClinVar annotations. So I am wondering how can we update the ClinVar matrix table for local installations?

Many thanks for your help, appreciate it!

Anjali

[hanars]

You are correct that issue arises from using an older clinvar matrix table. To fix it, you first need to download the updated table- this will depend on whether or not you are using google cloud or not:

To local files:

docker-compose exec pipeline-runner rm -rf "/seqr-reference-data/GRCh${BUILD_VERSION}/clinvar.GRCh${BUILD_VERSION}.ht"
docker-compose exec pipeline-runner gsutil -m rsync -r "gs://seqr-reference-data/GRCh${BUILD_VERSION}/clinvar/clinvar.GRCh${BUILD_VERSION}.ht" "/seqr-reference-data/GRCh${BUILD_VERSION}/clinvar.GRCh${BUILD_VERSION}.ht"

For GCP/dataproc:

gsutil rm -r "${GS_BUCKET}/reference_data/GRCh${BUILD_VERSION}/clinvar.GRCh${BUILD_VERSION}.ht"
gsutil rsync -r "gs://seqr-reference-data/GRCh${BUILD_VERSION}/clinvar/clinvar.GRCh${BUILD_VERSION}.ht" "${GS_BUCKET}/reference_data/GRCh${BUILD_VERSION}/clinvar.GRCh${BUILD_VERSION}.ht"

You will then need to re-load your projects and add the new indices to seqr in order to get the updated annotations in your search results

[hanars]

So other than clinvar, everything else annotated during data loading is contained in a single combined_reference_data table. You can update it using the same instructions as above for updating clinvar, except instead of clinvar.GRCh${BUILD_VERSION}.ht as the file path use combined_reference_data_grch${BUILD_VERSION}.ht

[anjalijain22]

That makes sense! We'll update that on our end as well. How often are these hail tables updated on your end (monthly, once every 6 months, yearly)?

Also going back to the original issue, we updated the ClinVar Hail table and now the variant is correctly annotated as likely benign :) Is it possible to add the update commands for the reference databases in the README for local installs? I believe this will be very helpful.

[anjalijain22]

Another thing that I just thought of is - in the event that we do reload data for certain projects for reanalysis, we would lose all the previous variant tags/notes etc done on the previous data loaded for that family, correct?

[hanars]

We update the annotation tables roughly every 6 months, or whenever we have a new annotation we want to add (maybe once or twice a year).

in the event that we do reload data for certain projects for reanalysis, we would lose all the previous variant tags/notes etc done on the previous data loaded for that family, correct?

That is incorrect. Tags and notes will remain! seqr will attempt to update the variant to use the latest annotations, so for example if you had previously tagged that "pathogenic" variant, after reloading the project when you go to the saved variant page the variant will be there with all the tags and notes but should be showing as likely benign now

[anjalijain22]

Oh I see, that's great to know!!

Thank you so much Hana for all your help with this, much appreciated!