Genome assembly -hiseq illumina data

[Bgansw]

Thank you Koadman and vamosia. Yes, I'll reduce the dept and try again.

@Koadman: I've downloaded your A5 pipe line. Thanks. I'll keep you'll updated with the result.

Bgansw

[vamosia]

It seems to me that your depth is too high. Since errors accumulate, the more depth means more chances of introducing errors, hence breaking the assembly. I would suggest try lowing the depth and see if there is any improvements.

[koadman]

Getting a good assembly can be surprisingly challenging, even for bacteria. Downsampling your data is probably a good thing to keep compute requirements manageable but may cause you to lose coverage in regions of extreme GC content, resulting in a more fragmented assembly.

Our group has had good luck with IDBA, and we have cobbled together an assembly pipeline called A5 that you could try. The A5 pipeline is supposed to be able to do bacterial-size assemblies directly from sequencer output by automating the adapter trimming, quality filtering, error correction, contigging, scaffolding, and some assembly QC. See here for more info: http://code.google.com/p/ngopt/

It's still very much beta, and we could really use some feedback on whether it's working outside our lab. In our hands it's generating pretty good assemblies, as measured by comparison to a reference genome. On your data you would almost certainly have to downsample to get it to work with the A5 pipeline, unless you have some big iron to run it on.

[Bgansw]

Hi,

I'm new to genome assembly, and am having problems.

I will be grateful for tips on how to assemble a bacterial genome.

I have 100bp hiseq illumina data. Genome size is ~4MB.

I have been using Dynamic Trim, then LengthSort, then velvet or soapdenovo.

The problem is that after running velvet, I get an N50 of 1Mb , but the total size of the genome come to 30MB, over 6 times more than it should.

A friend suggested using a subsection of the data I have (cause I have 350x coverage for one of my genomes and 3048x coverage for another). I've even tried that but the total size of the genome is still more than it should be.

Any help would be greatly appreciated!!