CHEVY CHASE, Maryland, June 6 – Scientists from the Human Genome Project and Celera Genomics left their long-standing rivalry behind Wednesday to participate in the Joint Genome Sequencing Workshop, a meeting designed to seek consensus on the best approach to genome sequencing.
“The goal is to try to look forward and understand what common ground we have, and how to best move forward into the future,” said Gene Myers, Celera’s vice president for informatics research.
In what appeared to be a move to offset any tension, researchers arriving at the meeting, which was held in neutral territory on the bucolic campus of the Howard Hughes Medical Institute, received nametags that conspicuously omitted organizational affiliations. The meeting was marked by an atmosphere that was collegial, yet cautious.
Gerald Rubin of HHMI, welcomed the group and reminded participants of a similar meeting held at that location earlier in the genome sequencing effort. At that time, tensions were set aside, albeit only temporarily, and a productive “lovefest” resulted.
Although some reseachers such as David Haussler of the University of California at Santa Cruz were less optimistic that a such a mood would prevail, they seemed committed to a truce.
"We're looking forward to a day of open and frank discussion. I wouldn't anticipate a lovefest at this point, but let's try," Haussler said.
In fact, most of the 70 or so researchers at the workshop came out in support of a combination of approaches, noting that some of the thorniest problems in finishing the human genome are best handled using a combination of the two projects' approaches.
Evan Eichler of Case Western Reserve University discussed using random reads from Celera's whole genome shotgun method as well as the public ordered data for studying large duplicated sequences in the human genome. In addition, different organisms' genomes might be most effectively sequenced using one method or the other, depending on the number and type of sequence duplications, GC-rich regions, and other factors, Eichler said.
As priorities are set for new sequencing efforts, it might be worthwhile to do shallow whole genome shotgun pilot surveys to determine which approach to take, he added. Cooperation between the camps would allow such decisions to be made scientifically rather than politically. "It's a pleasure to be here in this new era of detente between the public and private sector," Eichler said.
Of course, each approach still has its advocates, but they seem to be moving closer together, with most current sequencing projects employing some combination of BAC clones and the whole genome shotgun approach.
Jim Kent of University of California Santa Cruz cited the advantages of the HGP's hierarchical approach, including minimizing the confounding effects of repeating sequences, having a single haplotype within a BAC, and being able to easily distribute work across a number of centers. However, one way to optimize the hierarchical approach, he acknowledged, was to use the shallow initial whole genome shotgun concurrently with mapping.
Celera's Granger Sutton described the company's effort to eliminate its dependence on contigs from the public database. Although the sequence Celera published in February did use the public data, a more recent Celera-only human assembly at 5x coverage yielded dependable and significantly larger scaffolds, Sutton said. He attributed the improved results to enhanced algorithms and elimination of some errors found in the public data. “But nobody's arguing that 5x is a finished sequence,” Sutton said.
Like Kent, he said that one way to improve his project's approach was to apply a feature of the other method, in this case looking at BAC ends.
Jim Mullikin of the Sanger Center said the ideal mix of approaches depends in part on whether or not the genome in question is considered to be that of a reference organism. For a reference organism, he recommended starting with a 3x-5x whole genome shotgun in addition to making multiple large-insert clone libraries and generating end sequences and physical maps. The whole genome shotgun reads could be used to investigate the repeat nature of the genome, and then the appropriate mix of whole genome and clone shotgun depth could be determined.
For some non-reference genomes, an unassembled 1x-3x whole genome shotgun might be sufficient, he said. One difficulty is determining how to define a reference genome, as every organism has its advocates.
Jim Weber, a human geneticist at Marshfield Medical Institute, emphasized the additional depth required in the human genome sequence. He stressed the need to develop an efficient way to find and characterize all the common polymorphisms in the human genome, not only SNPs but also longer-range polymorphisms and insertion/deletions.
Medical research depends on understanding these polymorphisms, he said, and therefore it is important to do raw reads for more individuals than is necessary for other organisms.
"The finished sequence is not going to be the end. It's just the beginning," he said.