Researchers at Cold Spring Harbor Laboratory have developed a new way to capture and sequence fragments of a genome, which they plan to use in resequencing studies of schizophrenia.
The method, published online in Nature Genetics this week, couples NimbleGen microarrays with Solexa sequencing and is similar in principle to — though possibly less expensive than — a recently published paper by researchers at Baylor College of Medicine that combines NimbleGen arrays with 454 sequencing (see In Sequence 10/16/2007).
Also recently, two other genome-selection methods were published: one using NimbleGen arrays and resequencing chips, and the other a variation of molecular inversion probes and Solexa sequencing.
The challenge for studies using these approaches will be the extent of enrichment, the sequence coverage, and the data quality they yield, “since the ultimate goal is to find all the relevant genetic variations with a high confidence that the variants you find are real,” Michael Zwick, an assistant professor at Emory University Medical School and author of one of the earlier papers, told In Sequence by e-mail. “The next-generation sequencing technologies need to have high coverage ... to generate high quality data.”
Zwick, like the CSHL team, plans to couple NimbleGen arrays with Solexa sequencing in the future.
The CSHL researchers want to apply their method to study schizophrenia (see In Sequence 7/3/2007) with the ultimate goal of sequencing exons or other genomic regions from thousands of human samples, according to Dick McCombie, a professor at CSHL and senior author of the Nature Genetics article.
“We think [those are] the numbers you need in order to really impact that disease,” he told In Sequence this week.
He and his team developed the method primarily for this project, “although it clearly has implications for other projects as well,” McCombie said.
Cost was an important consideration, he said. Right now, sequencing an entire human genome at 6X coverage using the Illumina Genome Analyzer costs his lab about $50,000, McCombie said, and sequencing just exons will be significantly cheaper. “We are still trying to figure out exactly how much coverage we need,” he said.
Previously, Richard Gibbs at Baylor said that sequencing all human exons captured on NimbleGen arrays using 454’s platform costs about $100,000, but he pointed out that 454’s long reads “very well match exon reads.”
“At present, it is impossible to pick a winner because each paper presents different success metrics.”
All four recently published approaches have room for improvement, according to Maynard Olson, a professor in the department of genome science and medicine at the University of Washington. For example, “no one has yet shown satisfactory coverage of the targeted DNA,” he told In Sequence by e-mail last week. In addition, they all show biases towards certain captured fragments.
In a paper published last week in Nature Methods in which he commented on the three earlier reports, Olson wrote that at least for resequencing applications, selecting megabase-sized fractions of large genomes is “the central bottleneck.”
“At present, it is impossible to pick a winner because each paper presents different success metrics,” he stated.
Olson also wrote that repeat sequences in the genome “pose serious challenges for any targeting method that uses hybridization to capture short fragments of genomic DNA.”
PCR-based resequencing approaches, used in several recently published cancer exon studies (see GenomeWeb Daily News 9/12/2006 and In Sequence 3/13/2007), will be replaced by the new methods “if and only if these [new] methods deliver comparable false-positive and false-negative rates under realistic conditions of use,” Olson wrote.