variantvalidator check gives unexpected output

[seoanezonjic]

Hi @ielis
I'm using gpsea and it's an awesome library but I've a little problem when the gene variant data is not curated by myself in several projects. My problem is that in some cases, my collaborators give me malformed hgvs.c descriptions as the following:
NM_006772.3:c.359_
You can check in https://rest.variantvalidator.org/VariantValidator/variantvalidator/GRCh38/NM_006772.3:c.359_/NM_006772.3 the resulting json. This kind of json breaks the _extract_variant_coordinates method in VVHgvsVariantCoordinateFinder class that is called by the find_coordinates method. find_coordinates has a try exception but only triggers with two specific different exceptions. I generalized this to general Exception class to catch any problem with the query and keep patients with correct hgvs.c variants whereas the malformed ones are discarded.

Other suggestion is to add a verbose mode in VVHgvsVariantCoordinateFinder in order to output the rest url and the obtained response to debug these cases.

Thank you in advance
Pedro Seoane

[ielis]

Hello @seoanezonjic

thank you very much for the positive assessment and for pointing out the issue.

If I understand the nature of the issue correctly, GPSEA sends a malformed HGVS item NM_006772.3:c.359_ to Variant Validator API, the response complains about the incorrect format, which, in turn crashes GPSEA.

I will deploy a fix that will prevent the crash and report the invalid HGVS strings to the user (I think that's all that can be done).
So, the preprocessing should not crash with an input like NM_006772.3:c.359_.

If possible, please test if the new release (v0.9.9) works and let me know if the problems persist.

Thank you very much!

[seoanezonjic]

I checked the new version and it don't crash with malformed HGVS in my records. I assume that the warnings "Individual has no genotype data (variants) to work with. Add variants or remove the individual from the analysis" are due to the removing of bad variants. There is the possibility to give a message showing the detected malformed variants to keep track easily and report to the clinicians?
Thank you in advance
Pedro Seoane