samtools converting sam to bam filtering

[gringer]

Correct me if I am wrong, but the 'NM:i:XX' tag represents the number of mismatches in the read?

According to the SAM specification, NM should be used for edit distance (i.e. mismatches and INDELs combined), which seems to be the statistic that you want, assuming your aligner supports it.

FWIW, if you're just interested in mismatches, Bowtie2 uses a custom XM field for that. I'm not sure about what Shrimp does.

[dpryan]

Darn...I was trying to find the easy way out!

Correct me if I am wrong, but the 'NM:i:XX' tag represents the number of mismatches in the read?

Sorry to be the bearer of bad news. Yes, the NM tag is the one you want there.

[lre1234]

Darn...I was trying to find the easy way out!

Correct me if I am wrong, but the 'NM:i:XX' tag represents the number of mismatches in the read?

[dpryan]

There's no such flag in samtools, you'll have to code something up. The number of mismatches is pretty easy (you can actually use grep for that), but for the rest you'll have to parse the CIGAR string to get both the number of indels and their size.

[lre1234]

Do you know offhand what flag the reads that "have no more than 2 mismatches or insertions/deletions" have (in short, is shrimp2 calling then mapped or unmapped)? It may be easier to remap at a stringency that assigns the reads you want mapped as mapped, then you'll know that you are keeping those reads.

That's the problem. I didn't do the mapping, but rather was just given the mapped .sam file without much info, and am trying to figure out the number of mismatches for each read and further filter the file on those.

[ctseto]

Hi all,
I have been given some .sam files which were mapped using Shrimp2. I am now converting the file to a bam using samtools. My question is this: I know that I can use the '-F 4' flag to remove any of the unmapped reads. But I am also interested in keeping only those reads which have no more than 2 mismatches or insertions/deletions. When converting the file, is there an additional flag that will also do this type of filtering, such as an additional one of the hex flags?

Thanks

Do you know offhand what flag the reads that "have no more than 2 mismatches or insertions/deletions" have (in short, is shrimp2 calling then mapped or unmapped)? It may be easier to remap at a stringency that assigns the reads you want mapped as mapped, then you'll know that you are keeping those reads.

[lre1234]

Hi all,
I have been given some .sam files which were mapped using Shrimp2. I am now converting the file to a bam using samtools. My question is this: I know that I can use the '-F 4' flag to remove any of the unmapped reads. But I am also interested in keeping only those reads which have no more than 2 mismatches or insertions/deletions. When converting the file, is there an additional flag that will also do this type of filtering, such as an additional one of the hex flags?

Thanks