Merging paired end reads (R1 and R2 files)

[vectorborne5]

To anyone who may have dealt with Illumina MiSeq paired end reads: what are the best programs/scripts to use when merging the R1 and R2 (forward and reverse) read pairs into extended consensus..es (is there a plural for 'consensus')? I've tried four different program suites (pRESTO, flash, fastq-join, seqimp) and they all miss many, many read pairs that obviously overlap, sometimes even better than the reads that were merged. I sequenced adapter ligated microRNAs with barcodes. The MiSeq was readily able to identify the barcodes and eliminated most of the adapter sequence, and the read qualities are generally quite good, but I'm disturbed that my mergers aren't coming out as they should. I don't want to just take one of the files and accidentally identify a microRNA in a read that actually might have been a tRNA, or some other sequence type that I would have been able to eliminate by discovering that it was in fact a longer sequence after a proper read merger.

Any advice would be most appreciated!

My apologies if this question has been asked previously (tried searching, but nothing came up using various keyword combinations).

[JackieBadger]

FLASH generally works very well. However, it can miss and give you unexpected/false overlaps if your sequences are not properly trimmed.
MiSeq software for adapter trimming is very untrustworthy. You should re-Q your fastq files with adapter trimming turned off. Then FASTQC your PE reads to make sure that the base composition is exactly how you expected it.

I remember in my last FLASH run, there were some bases at the 5' of R1 that were not expected (~9 bases of an adapter or something..I forget), this screwed up the automated overlapping, which worked very well in previous runs.

So my advice would be to sanity check your sequences in relation to the expected sequence composition and length of overlapping region. You can also play around with the FLASH parameters.. I think you will be able to resolve it

[kcchan]

If you're running MiSeq Reporter 2.3 there's a new setting called "StitchReads" that can do this for you. Simply add StitchReads 1 to your [Settings] section of the sample sheet and you're good to go.

For more details, you can refer to the release notes for MiSeq 2.3.