Determine paired end overlapping

[chariko]

I have a paired end Illumina exome data set that might have overlapped at the ends. My fragment size has 105 bp. I aligned my samples with bwa and generated my bam files with samtools.
My question is:

Is there any way to determine how many reads overlapped?

How can I determine the distance between the paired ends (in case they didn´t overlap) or how much they overlapped?

If this percentage is high, I am thinking about reanalyzing my data generating fragments of 75 or 50bp. Do you think that’s correct? Which percentage could be the cut-off to consider it high?

Thanks

[volks]

in the SAM/BAM the TLEN column will tell you the template length. if it is smaller than 2*105bp you have overlapping ends (if there are no indels).

try this to see the template length distribution of the first million reads: samtools view PEalignment.bam | head -n 1000000 | cut -f 9 | sort -n | uniq -c

why would you want to generate fragments?

[chariko]

in the SAM/BAM the TLEN column will tell you the template length. if it is smaller than 2*105bp you have overlapping ends (if there are no indels).

try this to see the template length distribution of the first million reads: samtools view PEalignment.bam | head -n 1000000 | cut -f 9 | sort -n | uniq -c

why would you want to generate fragments?

Thanks for your answer, with the TLEN column I could manage it.

Regarding your question, the problem of having too much overlap is that I will miss the advantages of an paired end experiment that is detection of structural variants in the genome between the pairs for example. So if I generated 75 bases fragments I would have less overlapping. I know it's better if you work with longer reads but I thought this could be a solution. Also I think there are softwares that can get this info even if pairs overlap but I don´t know yet which of them. Do you have an idea?