Various assemblers and getting them to work

[avtsanger]

Hi

I have been having a go at running different assemblers on a variety of bacterial genomes. So far Velvet and Abyss have been very simple to use and have given me some useful output. I am however having trouble with some of the other assemblers and I was hoping someone out there might be able to help.

SOAPdenovo

After having obtained the version from the author that allows kmer values of greater than 31, I tried running it on a 2.5mb genome consisting of 54 bp Illumina reads. Using a kmer value of 47 (obtained using the velvet optimisation script) and run successfully using Velvet and Abyss, I tried to run SOAPdenovo. However, I immediately ran into trouble as the memory requirements exceeded 30gb and my job was cancelled automatically. Is this to be expected with higher kmer values and SOAP?

SSAKE

I typically use fastq files that have already been quality filtered. Is there a way of feeding these directly in to SSAKE. I may be being slow but it wouldn't accept fastq files...

ALLPATHS

Is it possible to run the assembler without a jumping library?

Any help would be gratefully appreciated.

[francesco.vezzi]

Hi,
SOAPdenovo: probably the data structures used by SOAPdenovo became too large with such big kmer, actually I'm a bit surprised that velvet is able to run with a kmer of size 47....
Another aspect is the meaning of a such large kmer.... with the de bruijn graph approach use a kmer of 47 means that you are looking for perfect overlaps of length 46, and with read length of 54 this is really too much because as you know illumina reads usually have errors at the end of the reads.

SSAKE:
it use the seed and extend paradigm that is unable to assemble even short repeats, I will not suggest it. I have not tested the latest version but the last time that I used it I found it results not really good

ALLPATHS:
my question is "is possible to run ALLPATHS?".... I never succeed in installing and running allpaths, I tried everything....
Every time that I spoke with other people that use de novo assembler they confirmed me that they were not able to use allpaths...

Francesco