The full complete day of Marco Island was overwhelming.  It opened with a general session in the morning, a diverse and fascinating set of talks, including:

• Jun Wang from BGI’s “Sequencing, Sequencing, Sequencing” in which he began to illustrate the quietly-growing powerhouse of sequencing in China.  Three campuses, 1,000 people, and who knows how many sequencers.  They completed their first whole genome (a Han Chinese) late last year, and are also sequencing the rice and panda genomes.  Dr. Wang’s explanation: “If it’s cute, sequence it.  Tastes good?  Sequence it.” Hilarious.
• Eddy Rubin from from the DOE-funded Joint Genome Institute on Genomics of Cellulosic Biofuels – a fascinating overview of how DOE is driving sequencing of switchgrass, poplar trees, and numerous other plants/organisms to extract sugar (and thence alcohol) from “biomass” that can be grown on marginal lands.
• Cancer talks – an overview of Canadian efforts and the ICGC by Tom Hudson, and Phil Stephens (in for A. Futreal) on structural variation in cancer.
• Comparative analysis of 2X mammalian genomes by Adam Siepel, a well-spoken individual who’s charged with correcting and analyzing the 40+ vertebrate genomes that have been sequenced to light coverage.

Strange Cats: Pacific Biosciences and Helicos

I was eager to hear presentations from the third-generation sequencing companies.  First came Pacific Biosciences, which made a big splash last year and seems to have been the golden child since.  Their SMRT-seq technology uses color fluore labeling of nucleotides that are read (at 3 bases/second) with a “zero mode waveguide.”  At the start of the talk, he fired off a real-time run that scrolled along colorfully at the bottom of his screen for the rest of the slides.  At present, they have 12 prototype instruments in house that run 3,000 reactions in parallel with average read lengths of 964 bp.  Since their recent proof-of-principle paper, they’ve also sequenced a 107-kb human BAC (to 68X) and a strain of E. coli (to 38X).  They claim accuracies of over 99%, though in the human BAC their technology missed 3/24 SNPs in non-repetitive regions and 7/20 SNPs in repetitive regions.  Even so, the data they showed was impressive, and I’m sure I’m not the only one eager for their early-access release planned for the second half of 2009.

Immediately following was Helicos, a company that I’d first heard of a few years go, but haven’t noticed much since.  Amusingly, they opened with the quote: “The rumors of our demise are greatly exaggerated.”  You have to appreciate candor like that!  Then came a not-too-subtle jab at the previous speaker: “Unlike technologies that go in a circle, we are going to show you true single-molecule data…”  They sequenced the canonical strain (N2-Bristol) of C. elegans to 27X coverage (7 channels, 88 million reads, with an average error rate of 3.4%).  The concerns I had were two-fold: first, error rates are fairly high, which will severely affect variant detection.  Second, their reads are only 33 bp long, and they readily admit that they’re not sure why.

Daniel MacArthur covered Complete Genomics

I chose a different session to attend, but Daniel MacArthur has a detailed profile of Complete Genomics’ presentation yesterday.  Though I doubt the big genome centers will have need of a sequencing-for-service-only company, the data they presented must have been impressive, because he was pretty fired up afterward.