Remapping reads

[ekageyama]

hello!
Currently I am working with a large set of STAR mapped reads, the output is in sam format. I would like to remap, but was wondering if its possible to do this using the SAM files without converting the sam files to fastq and then mapping. Any help would be appreciated.

[GenoMax]

You could use unix pipe (e.g. reformat.sh in=your.sam out=stdout.fq | some_other_aligner commands sequence input from stdin ). reformat.sh comes from BBMap suite. If you have paired-end reads then things may become more complicated.

Any particular reason you want to do this? Are you not happy with STAR's output?

[ekageyama]

Actually, I would prefer to stay with STAR, the reason is I need to map against a different version of the genome, and the fastq files were deleted for storage reasons, so I would like to avoid converting each sam to fastq all together, since im working with a fairly large dataset.

[GenoMax]

Then you could use STAR once again in place of "some_other_aligner" above.

[Dario1984]

If the new version of the genome is of higher quality and more complete, you would have lost all of those reads which did not map to the previously used genome but could map to the new genome, because the default value of --outSAMunmapped is None. I don't think that mapping again is worthwhile if you can't access all of the reads which were stored in the FASTQ files.