Cufflinks analysis for transcriptome assembly

Question

[qudrat]

Sometime while doing Cufflinks for transcriptome assembly it pauses at certain point of time. Does anyone have faced it, please help me

Answer 1

[Kevin Blighe]

For how long have you waited while it has been paused? It also pauses for me at a certain point during each run, but it always comes back provided that I have enough memory.

What specific command did you run/execute?

[qudrat]

I waited for at least 10 hour and i run the following command Cufflinks -o output -p 3 -M pseudogene.gtf accepted_hits.bam

[Kevin Blighe]

How much RAM have you got?

3 CPU cores is very little. In the past, I used as many as 32 CPU cores (human genome) and it still took time.

For a bacterial genome 1,000th the size of a human genome, it can still take 2-3 hours for Cufflinks to finish.

[qudrat]

Is there any way to split BAM file into two halves?

Answer 2

[grant.hovhannisyan]

This is most probably because of highly expressed transcripts (like rRNA or those coming from mtDNA). To quickly check it you can summarize reads by featurecounts and see which genes have very abnormally high depth. Then you can mask that genes using –mask-file option in cufflinks http://cole-trapnell-lab.github.io/cufflinks/cufflinks/.

[qudrat]

Actually i run the -mask-file for pseudogene

[grant.hovhannisyan]

Overexpressed genes are not necessarily pseudogenes. I my case out of 100mln reads 30mln reads were mapped to mitochondrial gene and cufflinks was getting stuck on that one. But after masking it the problem was solved.

[qudrat]

Thank you, let me check it