Strand bias reads

[Nino]

Hello All,

I would like to know if there is a way to filter out reads that are strand bias. When you upload a bam file to IGV you can see which strand position + or - the read is on and I would like to remove those that are like I said strand bias either complete on + or -. If this can be done through a vcf file.

Thanks,
Nino

[dpryan]

Do you mean that you want to remove a pair of reads if they both map to the same strand (rather than to opposite strands as is expected) or something else entirely? The only other reasonable interpretation of your question that I can come up with is that you've called SNPs on a dataset and found that some of the show significant strand bias and are trying to remedy this. While I've never sat and thought about it, I would suspect that there's no proper way to randomly remove reads to remove strand bias. If this bias is a concern, then just filter out SNPs with significant bias.

[Nino]

Yea sorry I do mean just SNPs I guess since the SNP is on the read and if the read is strand bias then you would filter out that SNP hence the read itself as well.

[m_two]

Varscan includes a filter to remove reads resulting from strand specific sequencing errors as well as mapping errors. Be careful how you implement a strand specific filter with capture data, the regions farthest from the probes tend to have single stranded coverage when the read length is shorter than the insert size.

http://tvap.genome.wustl.edu/tools/varscan/

[Nino]

Hey thanks m_two I will look into it.