Varscan For "Split" Regions In Rnaseq Reads

Question

[pulyakhina]

Hi everyone,

I'm using Varscan to call SNPs in my mpileup files generated from RNAseq data. The signs "<" and ">" in mpileup correspond for regions that fall within pieces of a split RNA reads, and these regions are calculated in the total coverage that is in the fourth column of an mpileup file. When I look at the Varscan vcf file, I see that the total coverage per in vcf files includes those < and > (I'm not talking about low quality bases, which are nicely removed by Varscan and I can see that the coverage in vcf file is sometimes smaller than the coverage in mpileup file).

However, I think that these "split" regions should not be considered as real coverage. Does anyone know whether I can tune Varscan in such a way that it takes this into account and does not include the coverage in those "split" regions into the SNP coverage?

Thanks in advance,

Irina

[Fabio Marroni]

Why don't you want to use split reads? However, if you don't want to use them, you can just remove them from the alignment file.

[pulyakhina]

I do want to use split reads. But when a read is split, the part in between the pieces of read that are actually mapped - this part is considered as a normal standard coverage in mpileup (marked with > or < sign) and I don't want this part to be considered as real coverage.

[Fabio Marroni]

Sorry, I didn't get it. I don't know how to help with this...

Answer 1

[dankoboldt]

Thank you for posting this question. VarScan does not count > or < bases for the purposes of variant calling. I just looked at the code to be certain.

It will report the SAMtools-reported depth, but it ignores < and > characters in the pileup.

If you have an example that seems to disagree, please send it my way!