TopHat disk space requirement?

[hr281]

Dear all,

i am trying to run TopHat on the RNA-seq datasets from Chris Burge's Lab (published in Nature 2008). However, TopHat seems to require gigantic amounts of disk space for its tmp file left_kept_reads.fq.candidate_hits.sam .
Just using 500.000 RNA-seq reads from the liver library (32bp long single end reads) it generated a 82GB left_kept_reads.fq.candidate_hits.sam file. Final output looks reasonable though.

Any ideas where the problem might lie ... or is that normal?

Many thanks for your help!!

Heather

[RiboFree]

Hi Heather,

Tophat does require such a huge disk space to keep temporary files, so it should be quite normal. The reason, I believe, is that it first aligns all the raw data by bowtie, then split unmapped reads into smaller pieces and align again. Finally, it'll delete temp files by default, and thus "output looks reasonable".

However, your 500k 1X32bp reads seems to generate too large files, compared with my experience(70M, 1X75bp), it just used up to ~100G. Did you use default parameters or specify your own?

Hope it helps,
-w