Take A Subset Of A Fastq Paired-End Sample

Question

[dfernan]

Hi,

I have two paired-end fastq compressed files coming from HiSeq RNA-SEq experiment, ie., pair.1.fastq.gz and pair.2.fastq.gz.

The files are very large so I wanted to just take a few million/thousand reads from each of them (by their respective pairs) and use that file for trying/debuuging purposes.

The results should be two paired-end files, i.e., pair.test.1.fastq.gz and pair.test.2.fastq.gz.

I'd be happy to hear some suggestions on how to do this or hear about tools available, thanks!

[Pierre Lindenbaum]

duplicate of

Selecting random pairs from fastq?

[dfernan]

thanks Pierre, I didn't realize someone else asked about it!

Answer 1

[Rahul Sharma]

Hi,

Assuming that the reads are in same order in both of the files. I would do like this:

zcat pair.1.fastq.gz | sed -n 1,4000000p > pair_1_millions.fastq
zcat pair.2.fastq.gz | sed -n 1,4000000p > pair_2_millions.fastq

Thanks,
Rahul

[dfernan]

Hi, thanks a lot, however, I am not sure if the reads are in the same order, I'd like to add that I am pairing them correctly...

Answer 2

[sebastian.gregoricchio]

I believe that a better option for paired-end data is to use fastq-sample from fastq-tools:

fastq-sample -n 5000000 pair_R1.fastq.gz pair_R2.fastq.gz -o pair_5M_R