Insert size for overlapped reads?

[serenaliao]

When using bismark, if we set -I(min insert size) as 20, will sam file exclude overlapped reads? In my case 70% of pe reads in sam file are overlapped, though I set -I 20.

Is this normal?

[chadn737]

The latest version of bismark uses bowtie2, which has options for dealing with overlapping reads and by default can map mates that overlap. I am not sure if bismark uses the default for this or not, or whether or not this is one of the options available to adjust in Bismark.

[serenaliao]

The latest version of bismark uses bowtie2, which has options for dealing with overlapping reads and by default can map mates that overlap. I am not sure if bismark uses the default for this or not, or whether or not this is one of the options available to adjust in Bismark.

A following question is If I use bismark methylation call extractor for the overlapped pair end reads(count the overlapped CpG twice), will the same reads actually report two CpG calling in the same site?(I want to take advantage of this to check the accuracy of methylation calling)

[dpryan]

A following question is If I use bismark methylation call extractor for the overlapped pair end reads(count the overlapped CpG twice), will the same reads actually report two CpG calling in the same site?(I want to take advantage of this to check the accuracy of methylation calling)

See the --no_overlap option. I recall that, by default, it will report both. With the --no_overlap option, I recall it reporting the call from the first read in a pair. I wrote my own version that checks to see if the calls in overlapping reads are equivalent and found that they differ in ~0.01% of cases in a dataset that I checked. BTW, the accuracy of methylation calls in this case will be purely a function of the sequencing accuracy.