contigs to fully assembled genome

[hhlee]

Hi all,
I've done a bit of E. coli resequencing but I am new to the art of de novo and could use some input on what strategies would give me a fully contiguous genome (and at what cost).

Trying to assemble a 5.2Mb bacteria (estimated from pulsed field) with 220bp SE MiSeq filtered reads (Q>=25) @ ~150x coverage. Velvet Optimiser tells me:

k=147
numContigs = 161
n50 = 81.6kb
longest contig = 279kb
num contigs > 1kb = 107
total bases in contigs = 5.1Mb (~98.9% covered)

I don't necessarily need a fully contiguous genome but what would it take? According to what I've read, a mate-paired run could help but it seems that I am at a point of diminishing returns.

Thanks for any suggestions!

[krobison]

I'd be interested why you chose 220bp SE over 150x2 PE.

Mate-pair libraries will likely drive up the continuity of your sequence -- you'll be able to use a scaffolder (such as SSPACE) to drive contigs into scaffolds -- though in many cases there will be gaps between contigs (and some of those gaps will be wrong and just cases where the two contigs overlap but by so little the scaffolder can't recognize it).

Another option would be Pacific Biosciences long reads. You'll need to use PacBioToCA or similar to merge them with your Illumina data, but these also can help. A number of cores or service providers offer this.

[hhlee]

User error is the reason why 2x150 PE reads weren't used

[Mark]

What is the purpose of your sequencing project? Are you sure a mapping or reference guided assembly wouldn't suffice?

[Gators]

Get 4 SMRT cells of PacBio Continuous Long Reads. That will give you about 70-80X coverage, and will bring your contigs to around 50, your max contig length to close to 1MB, N50 around 200K.

Should cost you less than 3000, including library prep, etc...i think