This is a test version of Biostars. For the public version, visit https://www.biostars.org.
The insert size distribution in RNA-seq fastp output shows 45-50% sequences of unknown length

Hey,

I am processing RNA-seq data from Illumina Hiseq 2500 to understand differential gene expression. I have performed QC and trimming of adapters in 'fastp' using the following command:

$ fastp -i SRR10439511_1.fastq.gz -I SRR10439511_2.fastq.gz -o SRR10439511_1_fastp.fastq.gz -O SRR10439511_2_fastp.fastq.gz --adapter_sequence AGATCGGAAGAGCACACGTCTGAACTCCAGTCA --adapter_sequence_r2 AGATCGGAAGAGCGTCGTGTAGGGAAAGAGTGT.

In the output file, the insert size distribution peaks at 255 and ~45% of the insert reads are of unknown length (Read length = 150 bp). Should I be worried? I presume that an overlapping read 1 and read 2 could be around 300 bp (2*150bp = 300bp). Insert seize distribution curve

It also mentions the following message: This estimation is based on paired-end overlap analysis, and there are 45.366302% reads found not overlapped. The nonoverlapped read pairs may have insert size <30 or >270, or contain too much sequencing errors to be detected as overlapped.

I am a newbie in sequencing analysis and would appreciate your help and provide more info if needed to help with this query.

read illumina rna-seq fastp length
• 3,682 views • link
updated by GenoMax GenoMax • written by Atreya • Atreya

1 answer

Should I be worried? I presume that an overlapping read 1 and read 2 could be around 300 bp (2*150bp = 300bp).

Probably not. Your data seems to have short inserts. Not much you can do about that now. See the explanation about insert sizes/read overlap here --> What is the difference between paired end reads and overlapping reads, and then why merge overlapping reads before assembly?

The nonoverlapped read pairs may have insert size <30

This may be likely explanation for the tail you see above. If you add a filter for minimum length (say 30) for reads remaining after trimming then some of that tail should go away. Really short reads will multi-map on the genome and are likely not going to be useful for counting.

• 0 views • link
written by GenoMax GenoMax

Thank you @genomax. The thread helped solve my query.

• 0 views • link
written by Atreya • Atreya

Log in to answer this question.

More posts like this