How to examine paired reads

[ramiro2k]

Hello, a beginner question, any help appreciated

We sequenced an organism (Illumina Hiseq) and would like now to examine how the reads are getting paired, just as a quality control look at how far the paired end are from each other, how many pairs there are.

Unfortunately, this organism does not have a reference genome, so initially our approach was to get a single gene, such as the rRNA gene fasta sequence for this organism, align as many of our sequenced reads as possible to that gene, and examine the distance between the pairs.

Does this approach make sense? What program can we use to align our reads to only a single gene.

Any other suggestions appreciated!

Thanks in advance,
Ramiro

[westerman]

I presume you are talking about an transcriptome project instead of a whole genome project thus avoiding the whole issue of intron/exons.

I wouldn't use rRNA -- too many duplicates -- but if that is all you have then, sure, use it. And it would be useful to have more than one gene to look at just for good statistics. In general you approach is valid for a first pass. As for the program to use, there are many. I'd try one of the short read aligners such as BWA, Bowtie2, SNAP, etc.

Also I suggest running FastQC to get a feel for the quality of the data.

[ramiro2k]

You are correct, it is a transcriptome project.

Thank you very much for your feedback. The reason I was asking about what program to use, is that BWA, Bowtie2, etc seem to be meant for long reference sequences (such as genomes, chromosomes) and not for a small gene of just a few hundred nucleotides in length. However, I will try.

I ran FastQC and the reads are of high quality.

The situation, on the bigger picture, is that the reads are not assembling, and one possibility is that the "pairing" has shifted for some reason (that what are being called "pairs" are not really pairs). I don't think this is true (for various reasons) but would like to verify it somehow before we rule it out and have that as a QC step for the future.

[westerman]

Certainly BWA, Bowtie2, SNAP are good for long references but they can also be used on short references. I use said programs to scan reads for matches to the rRNA database that we use. Said database has hundreds of short rRNA references but it could just as easily have one short reference. I also use the programs to scan my phiX database which has a single fairly short reference in it.