At the 10th anniversary of AGBT, new technology heralds exciting new biology.
By Kevin Davies
March 24, 2009 | MARCO ISLAND, Florida—While the scientific debut of Complete Genomics captured much of the attention at this year’s sold-out AGBT conference*, there were plentiful signs that next-generation sequencing technology is delivering much more than mass human genomes.
Opening the conference, Broad Institute director Eric Lander saluted the “intellectual and commercial ferment” in genome analysis that is not only generating unprecedented volumes of data but also spurring discovery. “It’s breathtaking, an entire community, academic and commercial, coming together,” raved Lander. Current next-generation sequencing instruments are churning out 2 billion bases (Gb) of sequence per day, numbers that will look puny next year. Illumina, Applied Biosystems, and Roche/454 all presented impressive data showing a doubling of throughput in 2008, with even more dramatic increases on tap.
Lander offered examples of how high-throughput sequencing is becoming a routine tool for tackling molecular biology problems. “For each disease, we’re going to have to sequence thousands of patients,” he said, to identify the rare causal variants. In February, Lander, John Rinn, and colleagues published a landmark discovery by studying chromatin state-maps to identify long intergenic non-coding RNAs (lincRNAs). Analysis of a telltale pattern of histone modifications—“If you see a K4 and a K36 [modification], it marks a gene”—delineates a staggering 1600 lincRNAs (up from the textbook tally of 12!). There is already evidence suggesting some have functional roles in gene regulation.
Les Biesecker (NHGRI) described ClinSeq, an ambitious study that launched two years ago to generate detailed sequence information on 400 candidate genes in 1000 subjects. “We can’t get patients through the clinical center as fast as they want to enroll,” said Biesecker. So far, his group has sequenced more than 825 megabases (Mb). In one patient with high cholesterol and a “stupendously high coronary calcium” level, the study revealed a stop mutation in the LDL receptor gene.
Wang Jun (Beijing Genomics Institute, Shenzen), said his institute is sequencing 20 Gb per day with a fleet of 18 Illumina GA II machines and a data center with 1000 CPUs and 1500 terabytes (TB) storage. Wang is selecting 100 species with little competition and a significant Chinese element, such as the giant panda, under assembly using in-house algorithms based on the Bruijn graph theory.
Rumors and Promises
According to CSO Steve Turner, Pacific Biosciences’ debut instrument won’t be available until late 2010, but there is healthy progress among the first prototypes. Last November, PacBio sequenced a 107-kb stretch of human DNA (in a bacterial artificial chromosome). The average read length was 446 bases, with some reads exceeding 2000 bases. Accuracy was 99.99% in the non-repeat regions. In January, PacBio sequenced Escherichia coli (38x coverage), finding just five discrepancies in the genome.
After a difficult year with job cuts, management reshuffles, and meager sales, Helicos Biosciences’ Bill Efcavitch declared, “The rumors of our demise are greatly exaggerated.” The HeliScope is producing up to 150 Mb/hour, with average read lengths of 30 bases. In a test sequence of the nematode, Efcavitch said the total error rate was 3.5%, and would move on to attempt a human sequence.
In a much anticipated talk from the CEO of Complete Genomics, Clifford Reid announced the release of his company’s first assembled human genome (250 Mb mappable reads). Reid said the quality—a discordant rate of 0.34%—was equal to the published African and Asian genomes. He anticipated achieving 200 Gb/run this summer, and triple that by the end of 2009, while still shooting for the $5000 genome later this year. A service model was the only way to go: “We’ll sequence, assemble, generate the variants, and send it back to you. 60,000 processors are going to light up,” he joked. (Editor’s Note: Reid keynotes Bio-IT World Expo on April 29, 2009.)
John Todd (Cambridge University) described progress in type 1 diabetes (T1D). Genome-wide association studies have implicated dozens of candidate genes, but which ones are causal? Deep sequencing of one such gene, IFIH1, uncovered four mutations, prompting follow-up studies in 10,000 patients. “The results were fantastic,” said Todd. IFIH1 turns out to confer protection when it harbors mutations. “When functional, you’re susceptible,” he said. More exciting still, the gene encodes an intracellular receptor for the coxsackie virus—a known risk factor in T1D.
Former George Church protégée Jay Shendure (Univ. Washington) used Agilent arrays to capture 25 Mb coding (exome) sequence of ten humans before sequencing. Included were two patients with a rare genetic disorder, Freeman-Sheldon syndrome. Shendure demonstrated that exome sequencing could pick up Mendelian mutations—21 genes contained novel variants, but only the mutation in MYH3 (the known disease gene) was predicted to be damaging.
The closing quartet of speakers began with Stanford Nobelist Andy Fire, who is using 454 technology to sequence immunoglobulin genes, despite admitting to “an F in immunology as a graduate student.” He hopes to use patients’ hypermutation status for a rapid prognostic cancer test. Bruce Budowle (FBI) said that biowarfare is nothing new, and dates back to the ancient Romans. His group uses SOLiD sequencing to sequence suspected anthrax and Yersinia strains. Len Pennachio described a wealth of data (see enhancer.lbl.gov) for identifying enhancer elements and characterizing their expression sites. Closing the conference was Penn State’s Stephan Schuster on ancient DNA and museomics. His group has sequenced the DNA of extinct and endangered species, notably the Tasmanian Devil, which is severely threatened by an infectious cancer outbreak. Schuster is sequencing Cedric—voted “Tasmanian of the Year” in 2008 for surviving test infections. The goal, said Schuster, is to use SNP information to “direct the breeding program and do pedigree selection in insurance populations.” In other words, increase the genetic variation to increase fitness. “We are racing to the finish line.”
Oxford’s Opening Statement
Oxford Nanopore Technologies recently published its first proof of principle of its label-free next-generation sequencing technology in Nature Nanotechnology. The paper*, by James Clarke and colleagues, shows that an engineered nanopore can discriminate between the four bases of DNA with remarkable specificity, and can even identify methylated C residues (“the fifth base”). Senior author is Hagan Bayley, Oxford University chemistry professor and co-founder of Oxford Nanopore.
Clarke’s team genetically engineered the alpha-hemolysin to covalently attach cyclodextrin—a washer that sits in the middle of the pore. By altering variables such as salt concentration, pH and temperature, Clarke could resolve the four bases to an accuracy of about 99.8%. The slight difference in size of the four bases results in a discrete blockage of the pore, which is read as a reproducible dip in the applied current from a baseline of around 60 picoamps to anywhere from 20-40 picoamps.
The authors conclude: “These advances represent the realization of a complete nanopore base detector, which, when combined with a compatible exonuclease DNA processing system, will provide the basis of a complete nanopore sequencer.”
That is the next step—to couple an exonuclease to the nanopore so that single bases are cleaved off a template DNA strand and their identity read off instantaneously as they funnel through the nanopore. The Nature Nanotechnology paper also shows that the nanopore can detect bases cleaved off DNA strands in solution by the enzyme.
*Clarke J. et al. “Continuous base identification for single-molecule nanopore DNA sequencing.” Nat Nanotech 2009
This article also appeared in the March-April 2009 issue of Bio-IT World Magazine.
Subscriptions are free for qualifying individuals. Apply today.