Tophat Time And Memory Requirement

Question

[Nicolas Rosewick]

Hi,

Anyone has an idea of memory and time requirement for tophat 1.4.1 for 76 paired-end reads (~60M reads) ?

Thanks,

N.

Answer 1

[Malachi Griffith]

TopHat version 1.x uses Bowtie version 1.x for the initial alignment phase. This step can take advantage of parallel processes, but the remaining steps are all performed using a single CPU. For a single Illumina HiSeq lane of ~150-300 paired-end 100-mers, it typically takes 2-5 days for alignment.

You may want to consider trying the recently released TopHat 2.0.0. This version incorporates many improvements and added functionality. Specifically relevant to your question is the use of Bowtie 2.x (faster than Bowtie 1.x for your read length). The release notes also state that many steps that were previously performed with only a single CPU will now take advantage of parallel processing. Both of these additions may result in significant run time improvements.

Memory usage has generally not been a problem for the TopHat portion of our expression pipeline...

Answer 2

[Sukhi Singh]

It depends on the power of machine you use, I work on cluster, so its comparatively fast I am not sure, if there is a complete parallel version for tophat to utilize a multi-core machine. Check this and this answers, same question asked here. Here, they talk about the parallel version but I haven't checked it.

Cheers