how to randomly select 20m reads out of a FASTQ file

[angerusso]

Hi All,

I have an RNASeq data from which I need to randomly select 20 million reads, around 5 times. The whole file is about 200m reads.

What is a way to do this? Does anyone have a script to share? Thanks.

[Richard Finney]

Googling/DuckDucking might have turned up the answer you are looking for.

Regardless, check this thread : http://seqanswers.com/forums/showthread.php?t=16505

Are your reads paired ?

[dpryan]

There are a few ways, some mentioned on this site and some over on biostars. One of those ought to work for you.

[angerusso]

Yes my data is paried end. Another complication is that the two pairs are of unequal size.

du -s command gives:
*_R1* = 64850642
*_R2* = 48640554

Googling/DuckDucking might have turned up the answer you are looking for.

Regardless, check this thread : http://seqanswers.com/forums/showthread.php?t=16505

Are your reads paired ?

[dpryan]

You're going to want to resync them before you do anything else. Google "paired-end fastq sync" for a plethora of solutions.

[angerusso]

So I ran the following perl script:

https://bitbucket.org/jesseerdmann/fairview-galaxy-dist/raw/8e6e7a346284c7c261c951998b0138e82f540f43/tools/msi/pe-sync-2-files.pl

and it says: "passed full check" using "quick" which means the two files are in SYNC.

"QUICK CHECK enabled
Casava 1.8 read id style
PASSED full check"

But before I use random selection of reads from the two files, following your google links, shouldn't I make them equal size? As R1 is bigger than R2, even through they are in sync, I assume they are in SYNC only for the reads size that's common between them. Am I right?

You're going to want to resync them before you do anything else. Google "paired-end fastq sync" for a plethora of solutions.

[dpryan]

I've never seen that perl script, so I can't say that it works correctly. If you follow the instructions from this thread on biostars (Pierre's comment first, followed by Steffi's), you'll get two synchronized files of the same size.

[dpryan]

I'll add, this sort of different number of reads in paired-end files issue usually only crops up when mates from a pair are trimmed separately. If that's the case here and you're the one that did the trimming, you're life will be easier if you use a different trimmer next time (trimmomatic and trim_galore are common choices).

[angerusso]

Ignore my previous msg. My files are same size when I used "du -b" command.

[dpryan]

As long as they're also the same when you use "wc -l" as well then things are OK.