Spikes in Insertsize too high

Question

[xpan]

Hi all,

In the insertsize plot which is produced by [picard CollectInsertSizeMetrics], there are two very high spikes on the posiiton 151 and position 528.

The analysis process:

1.[ trim_galore ]:Filter,and get cleandata;

2.[ Bowtie2 ]:with parameter “--very-sensitive], get bam file;

3.[samtools/picard/bedtools]: Filter flag 2304 , reads marked duplicate with picard and filter Blacklist region, finally get filted bam file;

I try to use bam file which is only after aligned with bowtie2, their insertsize plot also has the high spikes.

I noticed one another similar question, but the spikes are abnormally high in my case.

How can I fix this problem or What is the cause of this problem? Thanks for help.

[ATpoint]

Cannot explain where this comes from but you have a perfectly fine nucleosomal pattern. Continue with your analysis and ignore these little spikes. It is probably not even worth thinking about.

[xpan]

Thank you, I update the analysis pipline that it can solve the spike on position 500+. Bowtie2 has the parameter "-X"[default: 500bp], then I set it as 2000bp. This cause the flag of reads which insert size lenght larger than 500 will be from 97/145/81/161 to 99/147/83/163.But I don know the reason, I will check the manual of Picard.If someone know the reason, please tell me. Thanks.

[ATpoint]

That parameter only will render these reads as properly-paired but that should not have influence on the insert calculation, right? You really should not worry too much, it is just a few reads that have this insert size. Only spend time to investigate if you run into problems during downstream analysis.

[Yingzi]

Hi, did you find out the reason? I am having the same >500bp spike problem. I tried bowtie2 -X 2000 but it didn't help. Any info would help. Many thanks.

[ATpoint]

In all likelihood this is some edge case with the aligner. If these spikes were "real" one would see them in the library QC with the bioanalyzer/tapestation.