Extract random reads

[radclifr]

I'd like to create a small file of reads randomly taken from a much larger file; e.g., from ~200 million to ~10 million or less. The reason is that I just want to play with different mapping/trimming/etc parameters without waiting a day or two for results. I want it to be random to more or less represent the nature of the full file.

Does anyone know of any tools that can do such a thing? The original file can be BAM, SAM, or FASTQ.

Thanks,
Richard Radcliffe

[Richard Finney]

This is often asked.

Check out previous discussions :

http://seqanswers.com/forums/archive/index.php/t-16845.html

http://seqanswers.com/forums/archive/index.php/t-16505.html

[oyvindbusk]

Hi Richard.
You could easily extract random lines from a file using the e.g. the rand() function in perl.
-Ø

[mbblack]

The latest release of Samtools will do this with a simple switch

Samtools view -s input.bam > output.bam

where -s is followed by some fractional value between 0.00 and 0.99

eg. to pull out a random set of 25% of the reads in a bam

samtools view -s 0.25 in.bam > out.bam

[radclifr]

Perfect -- thanks. We had an older version of samtools...