PacBio zips out Haitian cholera genomes

[krobison]

Pacific Biosciences and a team from Harvard have sequenced 2 isolates from the Haitian cholera outbreak plus 3 reference strains in amazing time -- just over a month from deciding to collaborate to publication in a top journal! Actual sequence generation was 2 days after receipt of the bacterial genomic DNA.

Good article in BioIT World with the backstory, and currently free on New England Journal of Medicine with the science (paper; supplement with detailed methods).

[ECO]

Thanks for posting this Keith.

[Geneus]

Pacific Biosciences and a team from Harvard have sequenced 2 isolates from the Haitian cholera outbreak plus 3 reference strains in amazing time -- just over a month from deciding to collaborate to publication in a top journal! Actual sequence generation was 2 days after receipt of the bacterial genomic DNA.

Good article in BioIT World with the backstory, and currently free on New England Journal of Medicine with the science (paper; supplement with detailed methods).

Wow...a little bit of a "bow shot" to Illumina and ABI...I guess it really works.

[microphobe]

I've been interested in the Pac Bio system because of the long reported read lengths but I've found it difficult to find any reliable stats or raw trace data.

They quote some highly qualified stats with their system that sound good until you read the fine print, like claiming ~5kb read lengths but only when "strobing" (i.e. not actually recording the majority of the sequence.) and claiming >99.99% accuracy, but for the consensus sequence (i.e. the consensus after aligning dozens of reads, which doesn't say anything about the actual per-read accuracy).

Pac Bio seem to be very coy about releasing any meaningful stats or raw trace data... Are there any third parties out there who have any hands-on experience with this system and who can verify their claims?

EDIT: OK I just noticed the supplementary info for that paper, and Table S1 answered a few of my questions... single-pass accuracy was ~82%, mean read length ~1kb (I am assuming they didn't strobe for these reads) with best 5% of reads > 1.8-2.8 kb. The DNA was sheared to ~2kb so I guess they didn't resequence individual clones since as I understand it they would need ~500bp fragments to sequence twice with a read length of 1kb. All in all not bad for 2 day's work, but I'd still like to see some actual traces.

IMO, the system is still crying out for a photostable polymerase to allow real single-molecule consensus sequencing with long fragments. I hear they are working on it, definitely one to watch.

[Geneus]

My understanding is the early adopters of the PacBio systems have been under a type of "gag order" as part of the paid for beta program...or something to that effect.

[krobison]

If you sign up for their DevNet informatics site, they have some E.coli raw reads there -- but I can't claim to have actually played with them.

[ECO]

This link has been published before so I don't believe I'm crossing any lines...but the cholera dataset is available in the SRA here:

ftp://ftp.ncbi.nlm.nih.gov/sra/Submi...026/SRA026766/

[OlivierE]

In case anybody is interested, I posted a partial re-analysis of their dataset, with a few numbers and observations they don't directly provide in the NEJM paper.

http://oelemento.wordpress.com/2011/...uence-dataset/

[ECO]

In case anybody is interested, I posted a partial re-analysis of their dataset, with a few numbers and observations they don't directly provide in the NEJM paper.

http://oelemento.wordpress.com/2011/...uence-dataset/

Nice first post...!

[Joann]

Many thanks for your analysis, the data therein is very helpful/useful.