"No space left on device" after "Finished constructing BWT"

Question

[ari.nazarian]

This is my first time trying out a fastq -> VCF pipeline, so I don't have the experience to be able to determine if things are working as I'm chugging through each step (from trimming, to alignment, to variant calling).

I just tried to create an index using hg19.fa on my MacBook Air which has 4 gigs of RAM. I ran

bwa index hg19.fa

and after 695 iterations (and 3063 seconds) the program stops with:

[bwa_index] Update BWT... [fwrite] No space left on device.

I have about 2 gigs left on my hard drive, for what it's worth. But right now I'm just wondering what exactly was left to be done, since I got to the point where it says

Finished constructing BWT

Can I use the files (4 of them) that were created to align my paired-end fastq's? (From my understanding, there are supposed to be a lot more.) Also, if I clear some memory on my computer, can I resume this process (assuming that it in fact is not finished) from where it left off, using these 4 files? This is what was created so far:

hg19.fa.amb  hg19.fa.ann  hg19.fa.bwt  hg19.fa.pac

[Ketil]

I wouldn't trust them, bwa is perhaps more robust than your typical bioinformatics program, but chances are one or more of the files are incomplete, and that further processing will give you incomplete or wrong results - or if you are lucky, you will get an error.

Answer 1

[GenoMax]

Hate to say this but with 4GB of RAM you are not going to be able to use bwa on this machine. Ref this post by Heng Li who is the author of bwa.

[ari.nazarian]

That's rough.Thanks for the input, I thought maybe cause my FASTQ's were small files (5MB for each paired-end file) that this might work out.

Ok, well how about indexing itself? It seems as if indexing is still something I should be able to complete on my machine, no?

[GenoMax]

You could try getting a pre-made index from iGenomes site. The bundle is a big download though and contains several things (sequence/annotation) besides bwa index. 5MB sounds odd for a human sample unless you meant 5GB. In that case you really should look for an alternate computer/resource to do this alignment.

Answer 2

[mforde84]

Problem is you're trying to generate an index for a huge reference assembly on your laptop. Use AWS instead, it's cheap.

https://aws.amazon.com/ec2/pricing/on-demand/