Manta SV <DEL> with positive SVLEN

[davmlaw]

We use SVLEN as a way to enforce variant uniqueness

VCF format 4.4 SVLEN changes to be "length of structural variant" and becomes positive.

However, in 4.3 and earlier, SVLEN can be negative for DELs

A recently uploaded T2T VCF sorted_toref had positive SVLEN values for <DEL> variants. It had VCFv2

If you search for <DEL> on this page:

https://github.com/Illumina/manta/blob/master/docs/userGuide/README.md

You can see they are violating the specs

11      94975753        MantaDEL:0:1:2:0:0:0    T       <DEL>   .       PASS    END=94987865;SVTYPE=DEL;SVLEN=12112;IMPRECISE;CIPOS=-156,156;CIEND=-150,150        PR      161,13

Manta have archived their repo as read only and are no longer accepting bugs.

We should decide how to go for our DB - whether to stay in the new or old format.

Nobody really accepts the newer VCF format so we'd have to modify things eg for VEP or exporting etc

[davmlaw]

There were 7,824,469 variants, though we rmdup now

Todo: check if it works and delete del with positive svlen

[davmlaw]

Testing:

This should have no visible effect on Shariant, just standard VCF related things (imports, liftover) etc stil work

Will make a test case...