The best to process sample with 2 lanes of data?

[foxyg]

Hi, I have pair end illumina RNA seq data, each sample was ran on 2 lanes. What is the best way to process data of this kind? Do I just combine the rawdata or I process them separately?

[csoong]

I had a post discussing this you might find helpful. I dont know how to link it in here so you might search it in my profile.

Basically, it depends on the type of analysis you are doing.

[Jon_Keats]

Depends on your computation environment more than anything I'd say. Sometimes its nice to run the alignment in parallel at the same time, one lane at a time, or certain steps are to memory intensive for you enviroment if you put them all together. Sometimes worth testing both to see what works best for you. For instance if you use BWA to align and Samtools I'm not sure which is faster;
cat lanes together
bwa aln, sam generation
samtools sam-->bam, sort, index

or
run lanes separate (2 at a time if you can) with
bwa aln, sam generation
samtools sam-->bam, sort, merge, index

so in that example it is a question of if you can run in parallel or not?, do you have enough resources to sort all the lanes together, is samtools merge of parallel runs faster than aln all at once.

My guess is that running in parallel is much faster if you have the resources to do this, which most standard boxes purchased in the last year (dual hexa-core systems with at least 16Gb ram) should be able to do at least 2 lanes at a time easily