How to filter out chromosomes/specific regions from a BAM file?

[fcr]

Is there any way of filtering out specific regions from a BAM file. Let's say I want to exclude the sex chromosomes from some analyses, is there any way to do this from a processed BAM file?
Seems possible to do it by excluding those chromosomes from the reference fasta during the alignment. But does implies, ehem, starting the pipeline all over again!

Cheers,
Nando

[dariober]

Is there any way of filtering out specific regions from a BAM file. Let's say I want to exclude the sex chromosomes from some analyses, is there any way to do this from a processed BAM file?
Seems possible to do it by excluding those chromosomes from the reference fasta during the alignment. But does implies, ehem, starting the pipeline all over again!

Cheers,
Nando

I would make a file of regions you want to retain (e.g. a tab separated file with <chromosome> <start> <end> for all chromosomes but excluding, say, chrY and chrX). Then pass this file to samtools view.

Dario

[fcr]

Thanks Dario,

Ah, so it will work like this:

samtools view aln.bam regions_to_keep.tab

without the -R. Right??

Cheers,
Nando

[dariober]

Ahá damn simple, I guess this will work:

samtools view -R aln.bam regions_to_keep.tab

Thanks Dario!

...Hold on, I don't think -R is correct (I haven't tested it) I actually think you have to pass the file after the bam, without flag... (Sorry about the confusion)

[fcr]

Ok, no worries. thanks again for your answer.

[volks]

i dont think realigning without certain parts of the reference is a good idea, because some reads will instead be mapping elsewhere in the genome.

you could also use awk to filter:
samtools view -h aln.bam | awk '{if($3 != "chrX" && $3 != "chrY"){print $0}}' | samtools view -Sb - > aln.filter.bam

[lgmSeq]

To filter out specific regions from a BAM file, you could use the -U option of samtools view:

samtools view -b -L specificRegions.bed -U myFileWithoutSpecificRegions.bam myFile.bam > overlappingSpecificRegions.bam

-b Output in the BAM format
-L FILE Only output alignments overlapping the input BED FILE
-U FILE Write alignments that are not selected by the various filter options to FILE. When this option is used, all alignments intersecting the regions specified are written to either the output file or this file, but never both