BAM to FASTQ compression ratio

Question

[win]

Hi all,

I downloaded the Denisova BAM file from here.

Since this file is aligned to the 1000 genomes reference genome and I would like to align it to grch38 I started the following utility for BAM to FASTQ conversion http://gsl.hudsonalpha.org/information/software/bam2fastq

After the file converted the FASTQ file is about 180GB whereas the BAM file is 80GB.

My question is does this seem correct and is there a way to estimate the size of FASTQ from a BAM file?

If there is faster converter please could you share, thanks in advance.

Answer 1

[Devon Ryan]

BAM files are compressed, the fastq file that that outputs isn't, so yes, that's not unreasonable. Given the notice at the top of the bam2fastq page, I wouldn't be surprised if Picard is faster. It might also allow writing compressed files (no clue, I've never used its SamToFastq command).

[Pierre Lindenbaum]

FYI: SamToFastq

(...)
COMPRESSION_LEVEL=Integer     Compression level for all compressed files created (e.g. BAM and GELI).  Default value:5.

[Devon Ryan]

You saved me from looking at the documentation :)

Answer 2

[tszn1984]

use pipe '|' to avoid reading and writing huge files.

bam2fastx -q in.bam | bowtie -S /dev/stdin | samtools view -Sbh - >out.bam

[Devon Ryan]

I edited your post to make it a bit more correct. Note that if paired-end reads are being used here that that won't work (and don't forget orphaned reads...).