de novo assembly on 3.5Gbp data -- computational speed!!??

[bioinfosm]

I just posted a new version of euler-sr. It should run with less memory, so it will be less likely to start paging. The webpage is euler-assembler.ucsd.edu/portal

-mark

Are there some numbers on 'less' memory?

Thoughts on assembling human scale genomes? any appropriate tools?

[mchaisso]

I just posted a new version of euler-sr. It should run with less memory, so it will be less likely to start paging. The webpage is euler-assembler.ucsd.edu/portal

-mark

[jts]

Hello Jason,

Have you tried Abyss?

http://www.bcgsc.ca/platform/bioinfo/software/abyss

[jtjli]

just read the quickTutorial of Edena, it doesn't seem to work with paired end reads??

[baohua100]

maybe you can try Edena

De novo bacterial genome sequencing: millions of very short reads assembled on a desktop computer.
D. Hernandez, P. François, L. Farinelli, M. Osteras, and J. Schrenzel.
Genome Research. 18:802-809, 2008.

http://www.genomic.ch/edena.php

[jtjli]

yes i did.
i think that is more a problem of my 16GB ram limitation.
Nothing wrong with velvet (ssake is similar)

i'm just hoping someone can shed me some light on this

[bioinfosm]

i have used Velvet & SSAKE to build some contigs before and found that they can't take more than 10 million reads (75bp each) (it just gets really really slow beyond that)

but now i need to do a de novo assembly using much more than 10 mill reads (45mill reads/ 3.5Gbp). HOW am i going to do it? or should i just expect the program to finish in months instead in hours?

thx
jason

Wow, I did not know of such a limit to number of reads for velvet!
Did you check with velvet's author?

[jtjli]

Just illumina data
1 set of 75bp paired end reads
1 set of 35bp paired end reads
and 1 set of single end 75bp reads

[zee]

Are you working with paired-end reads or any usable mate-pair information?

[jtjli]

i have used Velvet & SSAKE to build some contigs before and found that they can't take more than 10 million reads (75bp each) (it just gets really really slow beyond that)

but now i need to do a de novo assembly using much more than 10 mill reads (45mill reads/ 3.5Gbp). HOW am i going to do it? or should i just expect the program to finish in months instead in hours?

thx
jason