Picard MarkDuplicates

[wangzkai]

This may be a naive question, but I was trying to figure out whether I should set "REMOVE_DUPLICATES" to true or false when using picard's "MarkDuplicates" to remove duplicate reads. Since I want to subsequently call variants using samtools pileup, I am not sure whether samtools pileup will then remove from consideration these duplicate reads that are marked by flags when it calls SNPs.

By setting the "REMOVE_DUPLICATES=true", my understanding is that the duplicates read will not even be written to the output file, which sounds a bit safer ...

Thanks for any insight on this!

[nilshomer]

This may be a naive question, but I was trying to figure out whether I should set "REMOVE_DUPLICATES" to true or false when using picard's "MarkDuplicates" to remove duplicate reads. Since I want to subsequently call variants using samtools pileup, I am not sure whether samtools pileup will then remove from consideration these duplicate reads that are marked by flags when it calls SNPs.

By setting the "REMOVE_DUPLICATES=true", my understanding is that the duplicates read will not even be written to the output file, which sounds a bit safer ...

Thanks for any insight on this!

Save the duplicates so you never lose any data. You can them use the "-m" option in 'samtools pileup' to filter reads based on their flag (including duplicates: 0x400). If you need to make sure that your "FLAG" is set correctly, see http://picard.sourceforge.net/explain-flags.html.

[xguo]

Save the duplicates so you never lose any data. You can them use the "-m" option in 'samtools pileup' to filter reads based on their flag (including duplicates: 0x400). If you need to make sure that your "FLAG" is set correctly, see http://picard.sourceforge.net/explain-flags.html.

Is the read flag included in pileup file? I'm thinking to call SNP using all reads and then filter them based on the number of non-duplicate reads supporting a SNP.