different read count of IGV and mpileup/varscan

[bongbimit]

Hi all,

I visualized the variant positions I got from varScan using IGV.
I saw that they show different coverage from what I have detected using varScan (usually the IGV have higher coverage for both variant and reference).
i also use -B in mpileup or --min-avg-qual 0 in varscan ( although my quality 's so good ~ 30 -> 40 phred score), but IGV alway have more read than IGV

Example
IGV
total count :235396
A:363(0%,213+,150-)
B:82094(44771+,37323-)
C:144(80+,64-)
D:152795(82683,+70112-)
N:0

Varscan
chr9 133738357 rs121913461 T C . PASS ADP=7996;WT=0;HET=1;HOM=0;NC=0;GENEINFO=ABL1:25;LSD=true;NSM=true;OM=true;OTHERKG=true;PM=true;REF=true;RS=121913461;RSPOS=133738357;S3D=true;SAO=3;SSR=0;VC=SNV;VP=0x050260000a01000002110120;WGT=1;dbSNPBuildID=133;EFF=NON_SYNONYMOUS_CODING(MODERATE|MISSENSE|Tac/Cac|Y253H|1130|ABL1||CODING|NM_005157.4|4|1),NON_SYNONYMOUS_CODING(MODERATE|MISSENSE|Tac/Cac|Y272H|1149|ABL1||CODING|NM_007313.2|4|1) GT:GQ:SDPP:RD:AD:FREQ:PVAL:RBQ:ABQ:RDF:RDR:ADF:ADR 0/1:255:7996:7996:5248:2736:34.22%:0E0:35:36:4528:720:2343:393

-> i see DP is very smaller than IGV

i also read dkoboldt's blog about 5 things of mpileup
Thank in advance.

[swbarnes2]

The obvious answer is that they are applying different quality filters, so that one software is including more reads than the other. You would have to look at what the default settings are for that software to know what quality filters are applied.

My guess is that mplieup is applying a hard limit of 8000 reads, like the first 8000 reads it sees at that locus, and then it's dropping a few for quality reasons. Varscan might be doing the same thing.

[bongbimit]

The obvious answer is that they are applying different quality filters, so that one software is including more reads than the other. You would have to look at what the default settings are for that software to know what quality filters are applied.

My guess is that mplieup is applying a hard limit of 8000 reads, like the first 8000 reads it sees at that locus, and then it's dropping a few for quality reasons. Varscan might be doing the same thing.

ok i fixed
i set -d1000000 in mpileup and my result was better(because of mpileup is limitted by 8000 read, but the DP or SDP value is not equal IGV result, i think i will try some parameter filter in varscan.

[bongbimit]

but i have another problem, some of my SNV have DP( or SDP) is not support by RD and AD

example:
chr9 133748238 . T C . PASS ADP=244208;WT=0;HET=0;HOM=1;NC=0 GT:GQ:SDPP:RD:AD:FREQ:PVAL:RBQ:ABQ:RDF:RDR:ADF:ADR 1/1:247:244208:244208:2:46:95.83%:1.9036E-25:35:33:1:1:25:21

chr9 133748239 . G C . PASS ADP=244208;WT=0;HET=0;HOM=1;NC=0 GT:GQ:SDPP:RD:AD:FREQ:PVAL:RBQ:ABQ:RDF:RDR:ADF:ADR 1/1:247:244208:244208:2:46:95.83%:1.9036E-25:37:34:1:1:25:21


you can see RD + AD is not equal SDP (or DP), how can i explain this problem?