Linux Distribution on next-gen analysis

[DZhang]

Hi,

Linux operating system is very popular in bioinformatics. Can everybody share what distribution you are using? What kind of software packages you are using and the pros and cons of that particular distribution.

I am planning to setup a Linux box for next-gen data analysis. So the feedback will be useful to me and other beginners.

Cheers,
Douglas

[lifeng.tian]

Our center purchased a Dell PowerEdge 1950 system with RedHat Enterprise, 32GB of RAM, 16 cores (Intel(R) Xeon(R) CPU X5570 @ 2.93GHz).

I installed ubuntu 10.04 on a iMac to access the server.

I am working on RNA-seq analysis, and I installed TopHat, SpliceMap, and BLAT.
1. Each BLAT process uses about 4G of RAM. Or you can try gfServer/gfClient which uses 1GB of RAM. I found BLAT is more sensitive for long reads (100nt, 125nt), however lower specificity than SpliceMap 3.3. If you are going to run BLAT, you'd better have more RAMs. On my server, I split the input to 7 parts and run it on 7 cores.

2. TopHat uses Bowtie, and SpliceMap supports Bowtie, too. Bowtie has much smaller RAM footprint, about 1GB for hg19. I usually run it on 12-14 cores and it is very fast. Here is part of the TopHat log, you can see that the alignment of 36M reads to hg19 only took 10 min*12=2 CPU hours.
[Fri Apr 30 02:58:20 2010] Checking reads
seed length: 100bp
format: fasta
[Fri Apr 30 03:01:24 2010] Mapping reads against hg19.masked with Bowtie
[Fri Apr 30 03:09:29 2010] Joining segment hits
Splitting reads into 4 segments
[Fri Apr 30 03:13:35 2010] Mapping reads against hg19.masked with Bowtie
[Fri Apr 30 03:23:36 2010] Mapping reads against hg19.masked with Bowtie
[Fri Apr 30 03:33:43 2010] Mapping reads against hg19.masked with Bowtie
[Fri Apr 30 03:43:42 2010] Mapping reads against hg19.masked with Bowtie
[Fri Apr 30 03:53:26 2010] Mapping reads against hg19.masked with Bowtie
[Fri Apr 30 04:00:53 2010] Joining segment hits
Splitting reads into 4 segments
[Fri Apr 30 04:04:59 2010] Mapping reads against hg19.masked with Bowtie
[Fri Apr 30 04:16:04 2010] Mapping reads against hg19.masked with Bowtie
[Fri Apr 30 04:27:11 2010] Mapping reads against hg19.masked with Bowtie
[Fri Apr 30 04:38:08 2010] Mapping reads against hg19.masked with Bowtie
[Fri Apr 30 04:48:36 2010] Searching for junctions via segment mapping
[Fri Apr 30 05:54:18 2010] Retrieving sequences for splices
[Fri Apr 30 05:56:24 2010] Indexing splices

Hope it helps.

[simonandrews]

This topic was covered fairly extensively in a previous thread