Input Error Using Varscan2.3

Question

[Jordan]

Hi,

I'm using Varscan2.3 to generate vcf files. I understand that we need to convert the BAM files to mpileup format first and then give it as an input. But I keep getting an error.

The command I used:

samtools mpileup -f ~/refs/human_g1k_v37.fasta file.bam | java -jar /varscan/VarScan.v2.3 mpileup2snp --output-vcf > raw.vcf

I get the following error:

Error: Invalid format for pileup at line 12
1       11829   A       0

There was a thread already (Varscan Error On Mpileup Input), but that didn't help me. Does anyone know how to resolve this?

Thanks!

[Liye Zhang]

Hi, I wonder if you have solved the problem, if so, would you mind giving me some suggestions on how to solve this?

[Jordan]

I did actually. I don't remember the exact command but, what I did was instead of piping it, I created mpileup's for both tumor and normal first. And then I used this command on the mpileup's

java -jar VarScan.jar somatic normal.pileup tumor.pileup --mpileup 1

This gives you the somatic variants directly.

Answer 1

[dankoboldt]

This is an issue due to the zero-depth lines reported by SAMtools mpileup, which VarScan does not expect, and should be fixed in v2.3.6. Please let me know if it's still an issue!

[Liye Zhang]

I resolved the problem, but this is something you might want to know.

I tried the samtools mpileup and pipe to Varscan.

It did not work when I used samtools 0.1.19 for Varscan2.2.11 ,Varscan2.3.5 and Varscan2.3.6.

But if I use the samtools 0.1.18, all versions worked.

Conclusion: the newer version of samtools may cause the problem.

Answer 2

[Fabio Marroni]

Try using pileup2snp instead of mpileup2snp (even if you use mpileup to build the pileup). This usually works for me.