Extracting Tophat Fusion Reads and converting to FASTQ

[medalofhonour]

Hello,

I have been trying convert extracted fusion evidence reads from a Tophat-aligned RNA-Seq BAM file (with "XF" tag) back to FASTQ format.

If I am not wrong, this is tricky because Tophat will have multiple lines of output for each fusion junction spanning read (where one segment of a single read maps to one location and the other segment to a different location on the reference) and Tophat will report one line per segment (in case of unique mapping).

If I grep out the fusion evidence reads using a simple "grep XF" and try to convert this to Fastq using Picard SamToFastq, I get an "Illegal mate" error, which is expected since the segments are so scattered across multiple lines.

Does anybody know a better way to do this ?

[dpryan]

You'll probably have to write a script to do this (even if Picard's SamToFastq can handle chimeric reads, tophat's output isn't in accordance to the SAM spec. (to be fair, tophat-fusion predates the change to the spec.), so it probably still wouldn't work).