Effect of removing secondary/supplementary alignments from BAM file?

Question

[cg1440]

Hello.

Is it a good practice to filter out supplementary/secondary alignments from a bam file before proceeding with variant calling (SNPs and INDELs) and obtaining the consensus genome? Or are such reads not taken into consideration by variant and consensus callers (specifically bcftools and iVar), therefore keeping or removing them would not have any effect on the quality of the obtained variants/genomes?

Thank you in advance.

EDIT: I also want to ask, in bcftools mpileup there's the option -A, --count-orphans Do not skip anomalous read pairs in variant calling. In your opinion, should I include these in variant calling, given that I have paired-end reads?

Answer 1

[desouzareis.r]

Hi,

Looks like supplementary reads are handled as normal reads and the behavior can be changed by using the options -ff and -rf in n bcftools mpileup (see here). The "anomalous read pairs" or "orphans" are those paired reads where the mate is not paired (see here).

[cg1440]

Is it better to exclude supplementary and include orphaned reads in variant calling? Is there any advantage for modifying the default behavior of mpileup regarding these 2 flags?

Also, thank you for the provided links, they're really helpful!

[desouzareis.r]

For snps and indels variant calling, I think it is better to exclude supplementary reads and orphaned reads could be included as it does not have significant influence on the result. I think the default behavior already meets what you need for variant calling. In general, I don't think it is a good practice to filter out from bam file the suplementary/secondary alignments, they should only be flaged and the tools for variant calling normally handle these reads well and some flags are very important for structural variants calling for example.