mapping reads to single human chromosome

[rodrigo.duarte88]

Dear colleagues,

I have started working with RNAseq very recently and got stuck with the following:

- Should I map my reads to the whole human genome or only to the chromosome I am interested?

I've seen mapping to a single chromosome can skew my data, but it took 2 days for tophat to map the reads of a single RNAseq library to a single chromosome (I am on an octa-core PC running Ubuntu).

Should I definitely buy access to a cluster or there are ways to over come this?

[GenoMax]

Sounds like you are not using multiple threads on your PC (-p option, not all parts of tophat are threaded so don't expect "n" fold speed-up).

Do you have an extraordinarily large input data set?

[rodrigo.duarte88]

I did use this option. I used all 8 cores of my PC, the machine (the cooling fan) sounded like a train for those 2 days. haha

Because our lab is interested in allele specific expression, we aimed for high depth of sequencing. Each fastq file generated (Forw and Rev) contains approx 170 million reads each.

I know in the long term using a cluster will be essential.. But now for primary analysis I am wishing I didn't need one..

[GenoMax]

If your primary objective of getting the mapping done is now over then you could move on to other analysis

If you expect to do this for multiple samples then using a more robust computational resource is a no-brainer. Using a single chromosome (or even the region where you gene of interest is) is a possibility but due to the short nature of illumina reads you may get false positive hits (to regions of common domains/repeats etc).