Linking non-overlapping paired end reads

[batwoman]

Hello,
I am trying to link non-overlapping paired end reads using Pear v0.9.8. The target amplicon length is 316bp and the read length is 151bp. Attached are my PrinSeq results for one Illumina MiSeq library; high quality, ~ 2 million reads. However, when I attempt to link forward and reverse reads, presumably using positional data, the majority of assembled reads are much shorter than 2 x 151 bp... but there should be zero overlap. I disabled the statistical test, have experimented with different settings; when I set the minimum assembled fragment length to 300bp, I get zero assembled fragments.

I am not a programmer and am new to using these data. Is there a simple script that I can use to link forward and reverse reads, and ideally place an 'N' other some other symbol at the join?

[mastal]

I have not used Pear. but I think it will only merge/assemble the reads if they overlap. And if they do overlap, then 300 is the max length you could get for an assembled fragment, rather than the minimum.

Have you done any QC/adapter removal with programs like FastQC/Trimmomatic that would show you if you have reads where the insert is shorter than 300 bp and your reads contain adapter sequences?

[Brian Bushnell]

Pear yields an incredibly high false-positive rate, and if you have non-overlapping reads, it will merge them incorrectly with short inserts like you have noticed. So, don't use it.

You can do the operation you are describing with the BBMap package:

fuse.sh in1=read1.fq in2=read2.fq out=fused.fq fusepairs pad=14

You can also use BBMerge to estimate the actual rate of overlapping reads.

P.S. You should probably reverse-complement read 2 first, which you can do with reformat.sh. And of course, the best approach would be to go back and resequence the library at 2x250bp. Why would you sequence 316bp amplicons at 2x150bp so they don't overlap?