[jvanleuven]

I will remove the GC bin in my future assemblies, thank you for the input. While I do not have to worry about accidental cocultures (none of the bacteria I work with are culturable), I do sometimes have a difficult time separating genomes by coverage because the variance in the contigs belonging to one genome can overlap the with the contigs belonging to another.

[Rockx]

For a denovo assembly, I guess you could use the GC filter to assemble a draft genome. You could then use this draft as a reference for future assemblies using contigs filtered using other methods, as you mentioned jvanleuven.

I'm new at all this too, so it's a good learning experience to try these different methods.

[Bgansw]

Tried A5 pipeline on Hiseq raw data:

Hi Koadman,

My apologies for the delay in getting back to you. Had to go overseas!

I tried the new A5 pipeline, and although it worked, I'm having the same problem, that I have with the other assembly methods. The genome size at the end of A5 is 11.3mb when it should be 4mb. I'm attaching the log file that qsub (cluster) generated.

Our samples are very hard to culture axenically.. so I guess thats what the probem is. Though I didnt have this problem for the last set of samples of the same species (used the GA2 for genome sequencing, and not the Hiseq).Do you have any suggestions on what approach I can take now? I was thinking of mapping the fasta output of A5 onto my reference genome using Bowtie or Soapaligner.

[koadman]

Great to hear that A5 assembled your genomes. Yes, sounds like you've got a coculture on your hands. I would be surprised to if this is a Hiseq vs. GA2 issue, unless your samples were multiplexed with other organisms and there were barcode read failures or the person who prepped the library and loaded the lane accidentally mixed libraries. Probably unlikely, and not really a Hiseq issue but a possible sequencing issue to rule out.

Mapping contigs back to a high quality reference is a good approach for separating the "contaminant" from the desired genome. You might still miss some small scaffolds that belong in your isolate. The best way to minimize that problem is to have the best possible scaffolds!

Might be interesting to find out what else is living in your sample! I've had pretty good luck in the past picking out the dominant organisms in a mixed sample using EMIRGE, a 16s-based short read taxonomy analysis pipeline. It might tell you whether the contaminants are things you expect to be living in the environment with your desired organism or things that came from somewhere else in the lab. You might also try metAMOS on the raw reads or even a simple BLAST of the A5 scaffolds against nr and subsequent MEGAN.