Velvet misassembly with chromosome and plasmid

[hkg]

Hello,

I am doing de novo assembly of some E. coli genomes (~5 MB). The libraries were prepared using a Nextera XT kit, sequenced on an Illumina MiSeq, giving 150bp paired-end reads.

One of the genomes contains an already fully sequenced 90kB plasmid. The 'expected coverage' for this sample is only approximately 35X. I have tried to assemble this sample using Velvet with a wide range of kmer values (from 45-99).
When I aligned the assembled contigs back to the plasmid, I notice that the sections that hit the plasmid are generally fragmented amongst larger contigs (larger than the plasmid).

Does anybody have any advice they could share to avoid this plasmid and chromosome misassembly? Is it due to the low coverage (but I have seen this in a previous run with higher coverage)? Is it due to a possible difference in coverage (I'm not sure if the plasmid has a copy number greater than the genome)? Should I just move to a different assembler?

Any help and advice would be very much welcomed and appreciated!

Kind regards,
Heidi

[GenoMax]

Just for the genome that has known plasmid. How about separating the reads for the plasmid followed by doing the assembly for the genome and plasmid independently? You can use BBSplit to bin the reads: http://seqanswers.com/forums/showthread.php?t=41288

[GenoMax]

SPAdes is also good for microbial genome assemblies.

[hkg]

SPAdes is also good for microbial genome assemblies.

Thanks GenoMax. I will definitely try SPAdes - I was already intrigued by it.

And also thanks for the other suggestion, for binning. I think that would make a lot of sense for this one sample, but for all the other ones I don't know any of the plasmid sequences beforehand. I was hoping to use this sample as a bit of a test for how to process the rest of the samples.

[hkg]

Question regarding coverage in SPAdes

SPAdes is also good for microbial genome assemblies.

Dear GenoMax -- I have finally tried this sample with SPAdes, and it seems to work very well. Thank you!

However, I have a general question regarding it: What range should the "coverage" of the contigs/scaffolds be?

The plasmids contigs are around 8-9X, where the main chromosomal contigs are around 15X. Should the ranges be similar to in Velvet? (10-20X)