samtools mpileup only uses reads with IsProperPair set to 1?

[orionzhou]

Hi,

I have some questions about the samtools mpileup usage, what kind of filters does it apply before generating the consensus (or variant call)?

1. it seems all reads with the IsProperPair flag unset (0) are all discarded. Unfortunately, none of the reads in my BAM files has this flag set, so running samtools mpileup gives a blank output.

2. what about orphan reads (i.e., one read is mapped but not the mate, so it could not possibly be a "proper pair")? Yet, I'd like to incorporate info from these reads to make variant calling

3. I'd suppose mpileup will discard reads with 0 mapping quality, but just want to confirm

Thanks,

[Hit]

Does anyone have any opinion on these questions? I just posted a similar question in an independent thread before seeing this one:

http://seqanswers.com/forums/showthread.php?p=38526#post38526

I also observed the same number of SNPs from mpileup when not using "-q 1" to filter any reads and when using "-q 1" to filter alignments with mapping quality score 0.

Any suggestion would be appreciated!

[kopi-o]

I've run into this issue as well. I think it would be helpful if there was a option to allow non-properly paired reads in mpileup.

[lh3]

Use "-A" to override the default behaviour.