De novo assembler for 300 million Solexa reads

[zchou]

Can anyone tell how much RAM needed for 300M PE reads if we use soapdenovo? I tried soapdenovo and found 264G RAM cannot fill up the soapdenovo requirements.

[drio]

In my experience, SOAPdenovo better than ABySS for assembling large Solexa read set.
I successfully completed assembly using SOAPdenovo. :-)

what genome?

[odysseus]

In my experience, SOAPdenovo better than ABySS for assembling large Solexa read set.
I successfully completed assembly using SOAPdenovo. :-)

[jgrundstad]

How was your experience with ABySS?

Thanks,
Jason

[odysseus]

Thanks for all the comments everyone. I'm going to use ABySS.

[natstreet]

Sorry - I forgot to say that curtain only works if you have a reference as a starting point. I don't have hands-on experience with it - I just came across it and thought it might be worth a pointer.

Personally, I would give ABySS a try as a starting point.

[francesco.vezzi]

Hi natstreet
there is a thing that I don't understand about Curtain. It is a reference assisted assembler (it uses maq to align reads against a reference and then if improved the reference assembly) or after assembling with velvet or others it maps with maq on the contigs anc then uses the pair read information to improve the assembly?

Francesco

[natstreet]

I second giving ABySS a try. I find it's memory requirements are much more reasonable.

However, if you do want to get Velvet working you could look at Curtain, which might help you out.

[francesco.vezzi]

I think that a solution to your problem is use SOAPdenovo or ABySS that have lower memory requirements

Francesco

[Zigster]

you might need almost 10x that amount of memory to handle 300M reads
use a subset of the reads, trim aggressively, and use stringent kmer and cvCut settings

[odysseus]

Hi~ all ^^

I'm currently trying to assemble genome sequence and I have about 300 million Solexa reads (Paired-end; 220bp insert size).

When I using Velvet for assembling, I've got an error as below:
---------------------------------------------------------------------------------
velvetg: Can't calloc 18446744072010747658 InsertionMarkers totalling
18446744046528688288 bytes: Cannot allocate memory
Reading roadmap file ./Roadmaps
301083362 roadmaps reads
Creating insertion markers
---------------------------------------------------------------------------------

So I try to use other assemblers. If somebody tried to assemble paired-end Solexa reads and successfully completed, please tell me about the assembler and run command.

Server specification: 12 CPU, 72GB RAM

Thanks for any comments.