Picard-Tools Or Trimmomatic Flip Read From Fr Or Rf??

Question

[mengqingying]

I have some PE RNAseq reads contaminated with adapter sequences.

Before removing the adapters, i mapped the reads using bowtie2 -> samtools -> picard-tools "CollectInsertSizeMetrics.jar" and got histogram as shown in the first figure.

After removing the adapters with Trimmomatic, i again run bowtie2 -> samtools -> "CollectInsertSizeMetrics.jar" and got the 2nd histogram with many RF reads, as shown in blue color in the plot. Actually here i already set bowtie2 parameter --no-discordant to remove discordant reads.

I cannot figure out the reason there are so many reads flip their orientations from FR -> RF? I wonder if this change are caused by Trimmomatic, or picard? How this happened?! Any one can help?

[Dan D]

So your implication is that some reads changed their mapping orientation, but that's not what I see. Your second histogram contains more reads than your first, but it looks like the first histogram is fully duplicated in the second histogram. So I'd say that after trimming, more reads overall mapped, and picard thinks they're RF.

Having your full command lines for the bowtie and trimmomatic operations would be really helpful. If you didn't run trimmomatic in paired-end mode correctly, that could explain what you're seeing here.