Assembling pooled BACs from 454 data

[westerman]

We have a 454 titanium run of ~50 pooled BACs. Not bar-coded. Not paired-end. Two clonal lines. Previously mostly unsequenced genome. Genome is undoubtedly repetitive. BACs could overlap.

I am having trouble assembling the BACs. Newbler runs but then hangs in the 'deconvoluting step'. The TIGR EST clustering pipeline -- hey, I figured this was like an EST program only with bigger "ESTs" -- is throwing most of the reads into one contig even after masking out vector, adapters, etc. Of course ideally one would like to see 50 or so contigs which could then be assembled.

Does anyone have any papers to read or ideas on how to extract these BACs from the 350 Mbase dataset? I guess that basically I need a good clustering method. After that the assembly itself should be simple.

Thanks,
-- Rick

[mgenome]

How about dividing sequences into small groups?

After assembling the each group and gathering the contigs, you can assemble the whole contigs one more time.

Using more stringent criteria such as higher homology and longer mimium overlaps can be an another approach.

[westerman]

How about dividing sequences into small groups?

After assembling the each group and gathering the contigs, you can assemble the whole contigs one more time.

A good idea and one that I will try. If nothing else I might get to the repetitive parts of the BACs.

Using more stringent criteria such as higher homology and longer mimium overlaps can be an another approach.

Yes. I was running several of these clusters last night only to come back to work this morning and find that I was over my disk quota and that my programs crashed in mysterious ways. Who would have guessed that 250 GB would not be enough space.