Reverse engineering BAM files: BAM -> FASTQ

[gene coder]

How can I possibly extract the reads from a BAM file and put them into a FASTQ file for simulation (maq simutrain, then maq simulate)?

Should I just extract col. 1, 10 and 11 from a BAM file and put them in a text file along with a '+'? That is the output by read simulators.

But does it not happen that DNA sequences and base quality sequences are reversed and/or transformed depending on the direction and strand of reads relative to reference genomes? I would have to fix that too in that case.

[raonyguimaraes]

http://biostar.stackexchange.com/questions/771/converting-bam-to-fastq

[Richard Finney]

Check out http://seqanswers.com/forums/showthread.php?t=16395 for bampe2fq.c and bamse2fq.c for fast implementations of bam to fastq programs. It handles your concerns about reverse complimenting the sequence and reversing the quality string.

[gene coder]

Thanks. I found that SamToFastq in Picard did the job on a chromosome of NA12878 from the 1000 Genomes Project.

1. I separated SE reads and PE reads by library into separate BAM files using SamTools.
2. For PE reads, I also had to get rid of read-pairs that were unmapped (bit 3) or whose mate was unmapped (bit 4) or that were not properly aligned (bit 2).
3. I further separated the BAM files by read length.
4. Each BAM file now contained SE or PE reads only of the same read length and the same library.
5. Now I could run SamToFastq to convert a SAM file to a DAT file for use with MAQ simutrain.

Those were the steps to do to get what I wanted.