De novo assembly: raw data type & volume

[moinul]

Hi,
I'm trying to assemble 454 raw reads (gDNA) with Newbler 2.6. Can anyone tell me what is the maximum volume of raw data (nt) newbler could intake in one-step or incremental form (as it could assemble large genome of up to 3Gb in size) ? Also the proportion of shotgun and mate-paired reads we should use in order to have a better assembly ?

[bioBob]

I have done some assemblies with some pretty large data sets, in the 40-50 Gb range. With the large and het options, I can get an assembly, without, they simply never finish. By never, I mean, after 6 weeks of processing, no updates of the status files for several weeks. I did these on machine with 1 TB RAM. I tried various incremental assemblies and different parameters and essentially got to the same place as when I presented Newbler with all the data. I didn't see any improvements with the CIO options.

[moinul]

Thanks Bob........did you use '-m' option or others advanced options? And also trim the dataset? I had the trouble when tried to feed the trimmed and split 454 mate-paired reads. Because Newbler couldn't detect them as mate-paired though it is not seen for Illumina paired-end reads after trimming.

And for a large eukaryotic genome, say 3Gb in size, the 40-50Gb dataset you used covers only 16-17(x) of the whole genome that might not be quite enough whereas, for a 300Mb genome the figure reached upto 167(x)! So, is there any rule that what coverage we should initially use while trying to assembling a large genome?

[bioBob]

Hi,
yes, we do quality and contaminant trimming. Newbler looks for the linker, so if you mean you are splitting the reads and removing the linker, that doesn't work. Or at least, didn't last time I did that.
As for rules of thumb, I always refer to the Broad's guidelines. Which often don't work, but you have to start somewhere.
And yes, we did try the -m an other options. I probably tried about 20-30 different combinations.