Helicos Biosciences is reporting the re-sequencing of the M13 viral genome.
What’s really cool about this is that it’s sequencing-by-synthesis, requiring no amplification before sequence reading, so there’s no biasing of the populations of DNA fragments. Here’s a short overview of how their sequencing by synthesis works:
So it’s still a shotgun-style technique, and the problem has been getting long enough read lengths to assemble the pieces into a whole sequence. Think of it like a jigsaw puzzle. The smaller the pieces are, the more you have that look identical, and the harder it is to figure out where to put them. Given 3 x 109 bases in the human genome, you’d need a read length of about 17 base pairs to be certain of ending up with all different pieces. They report average read lengths of 23 bp, with each individual sequence chunk represented about 150 times, and every part of the genome represented. This allows them to keep the error rate low enough that they could sequence different strains of the virus, and reliably pick up the genomic differences between the two. The paper doesn’t say how long this takes, but their marketing material says the process takes only one day.
The technique as practiced by Helicos has one major drawback, however, that will limit how low it can take the cost/base. It requires very expensive optics, since you’re essentially doing FRET on an array of targets. The list price for their instrument is $1.3 Million, with a consumables price of $18,000/sample, placing it out of reach of many institutions.
Illumina, another company with a sequencing-by-synthesis application, gets around the expensive optics issue by doing a clever solid-phase amplification of their target strands, essentially growing little colonies of identical strands in situ. A sales rep told me they’re selling around $500K, but didn’t have details on the consumables cost.
Long-term, I’d expect the Polonator to be the most widely adopted platform, due to their aggressive pursuit of royalty-free technology and low instrument cost. There’s a good thread to follow here, if you’re interested.
23andme and Navigenics take note: They’re not offering a direct to consumer service, yet, but there’s no way a SNP scan can compete with a full sequence, once the cost comes down.
via HotCites, more on this story at in-sequence.
Thanks for the link Mr. Gunn! 😀
Couple comments on your post…
I’m not sure what the latest experiences are on read length needed for efficient de novo human sequencing, but it’s much much longer than 17 for humans, considering the number of repeats and similar regions.
I might be getting my providers/technologies mixed up, but AFAIK there has not been confirmation publicly that Helicos uses FRET. There was some initial discussion of it between a labeled template strand and the incorporated nucleotide (http://biotechnicalcurrency.blogspot.com/2007/07/helicos-part-ii_25.html), but no confirmation on helicosbio.com…I’m interested in this, as this type of FRET will limit their read length (versus say a Visigen polymerase-nuc FRET setup).
Last but not least, Illumina consumables are ~$2-3k per sample (so ~$25k for an 8 sample run.)
There is further (humorous?) discussion of this paper here on my site: http://seqanswers.com/forums/showthread.php?t=191
Thanks for your comments, Eric. In practice, longer reads than 17 are needed, and Helicos is getting around 23, which gets them close to where they need to be. I was simply showing the reasoning behind the number.
I don’t have any insider information that Helicos is actually using FRET, but from reading what’s been published leading up to this, that’s certainly what it sounds like. It’s also consistent with the exorbitant instrument prices they’re listing.
Thanks for your additional link, too. I meant to include it in the original post and lost it somehow in the plethora of tabs I had open.