Computer spec for RNA-seq analysis.

[hong_sunwoo]

Hi.
I am a newbie at RNA-seq data analysis and have a plan to perform RNA-seq from human cell line using Illumina platform.

I found that numbers of software such as tophat and cufflinks is suitable to my purpose.

Is it possible to execute these programs in personal computer?
What is minimal and recommended spec for those analysis?

from beginner.

[GKM]

Depends on what number of reads you're dealing with and how much resources your computer has. It would be helpful if you can provide that information.

What I can tell you is that TopHat will take a very long time if you don't use it multithreaded, and memory can bean issue if you want to run Cufflinks on a lot of reads (and sometimes for TopHat too), although the latest version of Cufflinks runs a lot more efficiently.

If you have a few hundred millions reads, it will be hard for sure

[hong_sunwoo]

Thanks GKM below is more specific information

Thanks GKM.

RNA-seq is peformed using Illumina Genome Analyzer IIe and 1 lane was used for each samples.

Our facility website tells as below
'Illumina/Solexa Genome Analyzer generates approximately 50 million raw reads of 36 nt length per full run'

I can not check exact reads, however, I think similar reads were obtained in my case.

Actually, I am planning to buy a new apple computer.
And the spec is dependent on your suggests.
I think that multithreaded is about cpu and memory is about ram size.
Would you mind suggest specific specs?

[GKM]

I usually run TopHat on at least 8 cores and cufflinks on 8-16, on your data it would probably take 7-8 hours for TopHat (maybe more, maybe less) and 1-2 for cufflinks with that kind of machine. As I said, cufflinks has gotten very efficient since the last version, I just ran it on 80M reads and it was taking less than a gigabyte RAM, so memory should not be an issue. So if you're willing to wait, you can probably run the pipeline on a personal computer, although it will be more unpleasant than running it on a cluster.

Also, if you have 36bp reads and not 75bp ones, you should supply TopHat with a junctions file.

[mrawlins]

We run bowtie on computers with 8 cores (2 quad-cores, I think) and 24 GB of RAM. That suits our purposes. I definitely wouldn't run any of these algorithms on a laptop, and never with less than 4 cores. If you don't mind waiting overnight for runs, a 6-8 core desktop should be fine. If you can wait over the weekend or a week you might be able to use a high-powered laptop, though I wouldn't recommend it.

We generally have multiplexed runs that we can split up, and run each one on a separate 8-core machine, which speeds things up considerably. That's a much bigger hardware investment, but it makes things run fast.