Compute inner mate distance for paired-end reads in a bam file

Question

[bioinfo8]

Hi,

I am interested to calculate inner mate distance for paired-end reads in many bam files i.e. distance from the right-most end of the left read to the left-most end of the right read and all the properly paired QC-passed reads that map in the bam file (ref: Calculating inner mate distance for paired reads)

I have used CollectInsertSizeMetrics.jar (ref: Picard MarkDuplicates error after using bamaddrg: "bin field of BAM record does not equal value computed..." ) and got the attached insert size histogram for one of my bam files (which includes properly-paired QC-passed reads, sorted and indexed). However, I am unable to understand the histogram properly.

In this post: Mystery: What is going on with this Insert Size Metrics Histogram, I can see both FR and RF in the histogram, but not in mine.

Please guide for the possible explanations.

Thanks!

[Devon Ryan]

The post you're linking to was asking why such an odd thing was being observed. Do you expect mate-pair style reads for some reason?

[bioinfo8]

I am interested to compute all parameters and alignment statistics for my aligned bam files (Whole exomes: comparative analysis ) and one of them was inner mate distance between paired reads. I was looking for the explanation of histogram and came across with the linked post. I assume FR=forward read and RF=reverse read, but why the program does not show RF and only FR. What is the significance of this?

[Devon Ryan]

FR would be the normal orientation for illumina reads, RF is the orientation for mate-pairs, where the reads point away from each other. Those should normally not exist unless you have mate-pair libraries (you don't).

Answer 1

[GenoMax]

You can calculate insert size distribution with BBMap following @Brian's answer in this thread: How to calculate the Average Insert Size after mapping the reads to the reference genome using BWA

Answer 2

[Devon Ryan]

In addition to BBMap and Picard, you can use bamPEFragmentSize from deepTools to conveniently get fragment size metrics.

[bioinfo8]

As per Paired-end read confusion - library, fragment or insert size?, will it calculate inner mate distance distribution and not insert size distribution?

[Brian Bushnell]

For the inner size, just subtract the read length from insert size. E.g.

Insert size = 400

Library = 2x150bp

Inner size = 400-2*150 = 100