This is a test version of Biostars. For the public version, visit https://www.biostars.org.
Extract reads where both pairs maps properly and none of the pairs are split from a bam file

Hi,

I want to extract only those reads from bam file in a given region where both pairs are mapped properly and are not split.

Currently I am doing:

samtools view -h -f 2 -F 4 input.bam "chr:start-stop" | awk '$6 !~ /S/ || $1 ~ /@/' | samtools view -bS - > output.proper.nosplit.bam

Where awk '$6 !~ /S/ || $1 ~ /@/' removes the split reads. However, I also want the properly mapped pair of the split read to be removed.

I tried to run samtools view -h -f 2 -F 4 output.proper.nosplit.bam > output2.sam, but it is still not removing the pair of the split read.

Any idea?

Thanks in advance

samtools mapping genome alignment sequencing
• 2,503 views • link
updated by Pierre Lindenbaum Pierre Lindenbaum • written by rimjhim.roy.ch rimjhim.roy.ch

1 answer

create a list of read-names that are not properly mapped or unmapped or contain a soft clip or a hard clip.

samtools view  -F 2  in.bam | cut -f 1 > out.1
samtools view  -f4  in.bam | cut -f 1 >> out.1
samtools view -F4 in.bam |  awk '$6 ~ /S/ || $6 ~ /H/' | cut -f1 >> out.1

sort / uniq the list , sort your bam on query-name and use the list with https://broadinstitute.github.io/picard/command-line-overview.html#FilterSamReads with FILTER=excludeReadList

• 0 views • link
updated by GenoMax GenoMax • written by Pierre Lindenbaum Pierre Lindenbaum

can we use 2048 flag (and/or 256) to filter out split reads from the region of interest?

• 0 views • link
written by cpad0112 cpad0112

Depends on which flag your aligner writes the split reads to. To remove just the secondary/supplementary alignments I would use the relevant SAM flag, to remove all records that are split reads I would check for the presence of the SA tag.

• 0 views • link
written by d-cameron d-cameron

thanks @d-cameron

• 0 views • link
written by cpad0112 cpad0112

Log in to answer this question.

More posts like this