entropy of fastq file

[papori]

Hi,
I want to measure the entropy of fastq/bam file.
The samples sequenced by illumina hiseq 2000 on human.
Any ideas how to do that , or any available tools?

Thanks,
Pap

[lindenb]

compress each read with the gzip algorithm and plot a distribution of the compressed lengths ?

Edit: I just wrote this for fun: https://github.com/lindenb/jvarkit/wiki/FastqEntropy

[Brian Bushnell]

If you use compression, you should first convert to fasta and strip out the headers. I actually have a program, bbmask, that calculates entropy, but it just uses that information to mask low-entropy sequences below a certain threshold. Though you can try it if you wish - it reports the total number of bases masked.

bbmask.sh in=reads.fq window=80 entropy=0.75

That will mask low-entropy sequences using a window length of 80 bases (which must be less than or equal to read length) and an entropy cutoff of 0.75. The higher the entropy cutoff (ranging from 0 to 1) the more bases will be masked. So, you can at least discover the fraction of bases that are below a certain entropy cutoff. If you want an output file then you can add the flag "out=masked.fq".