Calculation of supporting reads in pileup2indel is wrong
Variant detection in next-generation sequencing data
Brought to you by:
dkoboldt
Menu
▾
▴
#2 Calculation of supporting reads in pileup2indel is wrong
open
nobody
None
5
2010-10-12
2010-10-12
Anonymous
No
We encountered a wrong calculation of nucleotide frequencies by calling pileup2indel.
In our data we found a 100% deletion of 2 bases in IGV, but VarScan reported, that one read is supporting the reference, which is not true.
The VarSan output is:
chr23 14398626 C -TG 1 40 97,56% 1 2 38 33 0.98 1 1 0 1 25 15 DEL-2-TG
The associated pileup entry is:
chr23 14398626 C 40 ,-2tg,-2tg,-2tg.-2TG.-2TG.-2TG.-2TG.-2TG.-2TG,-2tg.-2TG.-2TG.-2TG.-2TG.-2TG.-2TG.-2TG.-2TG.-2TG.-2TG.-2TG.-2TG.-2TG,-2tg,-2tg,-2tg,-2tg,-2tg,-2tg,-2tg,-2tg,-2tg,-2tg,-2tg.-2TG.-2TG.-2TG.-2TG.-2TG.-2TG GGIGH#IFEH>@AEIIIIIHIHIBB#EE3BBB<#IIIIIH
I think there is a logical problem by parsing the samtools pileup (VarScan.java, Line: 473), because the "erroneously" appended periods or commas are added before indels, not after. So you have to check, before counting, whether the next base is a '-' or a '+' sign.
Greetings
Alexander Graf