AVGLEN slightly different from END-POS value

dnanexus-archive / parliament2

Runs a combination of tools to generate structural variant calls on whole-genome sequencing data

Apache License 2.0

102 stars 39 forks source link

AVGLEN slightly different from END-POS value #50

Open MaestSi opened 5 years ago

MaestSi commented 5 years ago

Dear Samantha, I have noticed that sometimes it may happen that AVGLEN value is slightly different to the value you would obtain evaluating END - POS. Is it right that, given a call in the Parliament2 VCF file:

AVGLEN represents the 'average' of the SVLEN predicted by single callers
END - POS represents the maximum size of the SV, obtained doing a union (instead of an average) of SVLEN predicted by single callers

Or is there another idea behind this? Thanks

slzarate commented 5 years ago

Hi @MaestSi, I believe what you are saying is correct. However, this information is generated by a tool we use called SURVIVOR (not Parliament2 itself). I am tagging @fritzsedlazeck, the author of the tool, to comment.

fritzsedlazeck commented 5 years ago

Hi Avglen is computed across all samples. Start and end represent a real breakpoints which is the closest to capture the consensus position . So that hopefully explains why they can be different. Thanks Fritz

MaestSi commented 5 years ago

Hi @fritzsedlazeck, Thank you for your answer. I am currently analysing a single-sample VCF, so I guess there is another level of averaging. Simone

fritzsedlazeck commented 5 years ago

Hi Simone, Yes it can of course also be that the callers dont agree with each other. It's the same principal. Cheers Fritz