How To Detect Chimeric Single End Reads With Galaxy?

Question

[magvazaryarvtre]

Hello,

I am currently trying to use a local instance of Galaxy to create a workflow to perform metatranscriptomic analysis (this is the first time that I use a workflow management system). For now, I am looking for a tool to detect chimeric reads among single-end reads ; does anybody what is the better tool for that?

Answer 1

[Pierre Lindenbaum]

the latest version of bwa (bwa-mem) can detect the chimeric reads ( see Can Tophat be used to find the virus-host junctions ? ) but I don't know it there's an instance of galaxy running that version.

[lh3]

bwa-sw also find chimeric reads, but for short reads, it is slower and less sensitive than bwa-mem.

[magvazaryarvtre]

Thanks, it looks like some bwa wrappers exists for Galaxy, but I didnt figured out yet what version of bwa they wrap. Btw, do you think it would be possible to perform chimeric reads detection by using bowtie2?

edit : the current (14/06/2013) bwa wrapper available in the Galaxy tool shed wraps bwa 0.5.9 and calls the bwa index command, so it is probably not suitable for chimeric reads detection

[lh3]

For short reads, you can use bowtie2. You can use the --local mode and ask it to report two best hits. However, bowtie2 is not really designed for chimeric reads. bwa-mem should work better (biased opinion of course). As to STAR, my limited experience is it has a high false positive rate. I do not know how much that matters with downstream analyses. There is also YAHA, but I have not tried myself.

[Mikael Huss]

I don't think Bowtie2 is really the right tool for this particular application because it does not use split read mapping. The STAR aligner (http://code.google.com/p/rna-star/) can do it and I have seen it included in some Galaxy installations (not the main one though, I think)