BWA alignments and time

[cggj]

Hi all,

I am aligning Illumina 75bp paired-end reads with BWA. I did this before and it worked just fine. Now I've been trying to align a new reads file and it works fine while I keep monitoring it, but it is the second time that it just stops and judging by the size of the file generated it didn't finish processing all the reads, and I can stay here forever making sure that it keeps running...

So, I have a few questions... 1. In average how long does BWA take to align let's say 1,000,000 75bp reads and 2. Does this look like a software problem or my server is killing the process at some point?

Thank you

[drio]

Is the bwa process doing anything CPU? I/O?

It should not take more than 5/10min.

[cggj]

Yes, I can see it processing and its running in the background. If I log out and in again and check the processes it keeps running but eventually it stops without finishing... It is taking right now ~15 mins for every 250,000 reads approx...

[lh3]

when the error rate is high, bwa is slow. you may also consider to apply -q20.

[bioinfosm]

How are people working off of bwa alignments to call snp/indels? I know of samtools and varscan, but people with experience on which to prefer and why...

In my experience, bwa/samtools was reporting many more events than maq followed by maq's snpfilter, which meant more false positives, which would be good to filter out.

[lh3]

The equivalence to maq+snpfilter is bwa+samtools+"*varfilter*".