toAmos Out of Memory!

[lmrodriguezr]

Hello everyone,

I'm analysing Solexa reads from a bacterial genome currently unavailable. As first step, of course, I'm trying to assemble the reads, and my pipeline includes AMOScmp as an important step in the assemblage. However, my Solexa library has more than 11 millions reads, with 36 bp each, and I just can't compile the AFG file for insufficient memory. I tried in many computers, and the best I reached was ~9 millions reads in a tmp file (using a machine with 16Gb in RAM), but I just can't get the work complete.

I have the same library in a variety of other formats, such as PRB (the original file), fastQ (Solexa and Sanger/Standard), Fasta/Qual, and even the WGS' FRG format. Unless the last one is pretty similar to AFG, I can't make toAmos convert FRG to AFG.

Does anyone know how to convert FRG (wgs-assembler) to AFG (AMOS) in a memory efficient way?, or otherwise, does anyone know how to merge two AFG files into a single one?

I tried to make a script for AFG merging, but I'm still having some issues with the format definition, such as "What does FRG's eid means?" or "How must I order the RED's frg?".

Thanks in advance!

[dan]

I'm not sure if AMOScmp is what you want if you are doing 'de-novo' assembly (you said "analysing Solexa reads from a bacterial genome currently unavailable" - do you mean you don't have the sequence i.e. you are doing 'de-novo' assembly?)

I think you may want to try an alternative assembly method.

Do you have the complete sequence of a close relative? Could anyone suggest a good hybrid strategy in this case?

I have been reading about SAM format for short reads, I think this may be what you need in this case rather than formats for WGS assembly of long reads.

HTH,
Dan.

[lmrodriguezr]

Hi Dan,

Yes, I have the complete sequence of a close relative. And yes, it's a hybrid strategy involving an AMOScmp run as single step of various re-assembling and filtering steps.

Thanks,
LRR

[anyone1985]

You can try some softwares like maq, Bowtie, which can map the short reads sequence to a reference genome. Maybe the result you get with such softwares will be thousands of contigs.