Genotype difference between Graphtyper & SURVIVOR

[Sreelekshmi-291]

Hello @hannespetur

We merged approximately 1000 samples using two different approaches: the SVIMMER + Graphtyper combo and SURVIVOR. After noticing some discrepancies between the outputs from these two methods, we extracted a single sample to investigate the differences more closely. The following is a snippet of what we observed,

here, "_G" is for the columns of Graphtyper & "_S" is for those of SURVIVOR.

As we see from the table above, for the variants called at the same POS, Graphtyper and SURVIVOR give 2 different Genotypes completely.
Also, We did not find any read supports in these locations in the sample's bam file.

What is particularly worrisome is that Graphtyper confidently calls a homozygous genotype at these positions, whereas SURVIVOR reports the genotype as missing.

Can you please provide some insights, that might help us figure out what is happening here?

Thank you

[hannespetur]

Hey. It is hard to understand something like this with only a small part of the output. It would be best if you could make a simple example of the problem that can be reproduced. You have no quality values in your table, what are you calling confident calls? I also don't see the genotyping model. Are these perhaps calls from the coverage model? in which case the support is the lack of read coverage in these locations.

Best,
Hannes