Strange coverage histogram of museum samples

[pitch_]

Hi everyone,

I had sequenced ancient museum samples on Illumina Hiseq 4000 platform. After running through typical pre-processing steps to generate a bam file (eg. cutadapt, bwa mem, mark duplicate, indel realignment). I then input the bam files into qualimap and obtained very strange coverage histograms (50x) (attached). I obtained coverage histograms with a similar pattern even after rescaling the bam file using mapdamage 2.0.

I am wondering what is the reason for this pattern in coverage?

Thanks in advance!

[luc]

Just a guess: paired-end reads of short fragments; thus they are mostly overlapping; resulting in counts as multiples of two?
More info would help.

[pitch_]

Just a guess: paired-end reads of short fragments; thus they are mostly overlapping; resulting in counts as multiples of two?
More info would help.

Hi Luc, that does sound like a plausible reason. What other info would help in confirming it?

[luc]

They are paired-end? Which read length?
You should be able to determine the insert sizes from the alignments (bam files).

[pitch_]

They are paired-end? Which read length?
You should be able to determine the insert sizes from the alignments (bam files).

They are paired-end reads with a read length of 150bp sequenced on illumina Hiseq4000 platform.

The insert size determined by qualimap for the 2 samples are
1st: 4194.82
2nd: 33,264.66
I am unsure how else to determine the insert size..

[SNPsaurus]

I think luc is correct. You can inspect the bam file to see if the paired reads map to the same location. You can also run bbmerge and see if the two reads overlap and what percent of the library does that.

[pitch_]

Hi all,

After checking the fragment size and mapping, I found that both of you were right, the coverage pattern is due to overlap of read 1 and read 2.

Thanks for the help!