Closed FelixKrueger closed 7 years ago
I have now tried to add functionality for the --dual_hybrid
mode to identify positions where both genomes had homozygous SNPs compared to the reference but where one strain did not pass the high confidence filters. Instead of making this a new option this is now the default behaviour since I believe this is the right thing to do. Addressed 210af817b2ca2c3681226896381a119630a7a6f9 and 1ab9048f92fb475eea68e59804085deb7be4d382.
In addition to high confidence homozygous SNP positions we also see some cases of low confidence no-SNP positions, such as this one: GT:GQ:DP:MQ0F:GP:PL:AN:MQ:DV:DP4:SP:SGB:PV4:FI 1/1:21:12:0:152,21,0:128,12,0:2:55:9:3,0,7,2:0:-0.662043:.:1 0/0:.:5:0:.,.,.:.,.,.:2:47:4:1,0,4,0:0:-0.556411:.:0
In line with only including high-confidence positions for the allele-specific analysis I have now added an additional check so that both FI
fields need to have passed the filter (i.e. FI=1) irrespective of the genotype (which may e.g. be 0/0, 0/1 or ./.). This addition requires some additional memory compared to the original version but will make the genome preparation more robust.
Addressed in c9688d92e7c543de262ce8eaef96b87a6b7585eb and 481a4605332d3580f231ad0cf8e6dc6f937b343d.
We have come across certain position in the genome where different strains appear to have the same SNP (indicated by the GT/genotype field), but one of the strains failed the FI/FILTER criterium (1 is PASS, 0 is FAIL). Here is an example:
GT:GQ:DP:MQ0F:GP:PL:AN:MQ:DV:DP4:SP:SGB:PV4:FI 1/1:22:6:0.166667:152,22,0:137,18,0:2:36:6:0,0,6,0:0:-0.616816:.:1 (129) 1/1:15:4:0:79,15,0:67,12,0:2:24:4:0,0,4,0:0:-0.556411:.:0 (Cast)
For single hybrid genomes we would include this position into the 129 genome (1/1 homozygous SNP, first line), but would ignore the position for the Cast genome (also 1/1 homozygous SNP, but failed the high confidence FI filter, second line). This seems like a reasonable approach.
For dual hybrid genomes such positions might be a problem though because when the 129 and Cast SNP lists are compared with each other it looks like there is now a SNP between 129 and Cast, even though there was evidence that the genotype was the same (1/1) in and Cast, only that it did not pass the threshold to count as high confidence SNP in Cast.
As a solution to this can we change the
SNPsplit genome preparation
to store theFI
value as well as theGT
genotype and only use the position for a dual-hybrid SNP list if the position was measured with high confidence (i.e.FI=1
) in both strains? Thanks to @nservant for helpful discussions in this regard.