National Institute of General Medical Sciences
National Institutes of Health
December 10-11, 2008
The PSI Steering Committee met as part of the PSI annual meeting, starting at the end of the first day and continuing until mid-morning of the second day. The Steering Committee consists of the fourteen PSI center directors, three members of the NIH staff, and five scientists not affiliated with the PSI, all of whom were present except Claire Ligett-Fraser, David Stuart and Yoshiyuki Yokoyama. Everyone assembled in this open meeting was welcomed into the discussions.
Much of the work of the Steering Committee is performed by six subcommittees. These subcommittees greatly facilitate coordination of PSI effort, and the impressive progress made in meeting goals, especially for structure production by the four large-scale centers, would not have happened without such coordination.
Target Selection. Subcommittee Chair Andrzej Joachimiak summarized the efforts by the large-scale centers on target selection, noting that emphasis is increasingly being moved to biological relevance from being mainly on structural coverage. John Moult presented an informative analysis by the Bioinformatics Group (BIG) of structural coverage of UniProt. He showed breakdowns by protein, by domain and by residue; coverages for selected model organisms; and the fractions contributed by the PSI. PSI makes dominating contributions in terms of novelty and leverage. Coverage at the 30% sequence identity level is at ~25% completion of structure space overall, higher for some organisms; but overall coverage rises to ~70% if analyzed at a threading level of similarity.
Goals and Milestones. Subcommittee Chair Guy Montelione related that the PSI-2 effort is on target to meet the goal established by the Subcommittees for PSI-2 to deposit 3000 structures into the PDB; the tally to date is 2100. Procedures are now in place for statistics on metrics defined by the Goals and Milestones document to be uploaded into the Knowledgebase (KB) on a frequent basis. Adam Godzik showed an evaluation of levels of coverage for various metrics that are tracked by KB.
Inter-Center Relationships. Subcommittee Chair Tom Terwilliger noted that efforts to share PSI-developed technologies have been successful, and that the KB now makes such dissemination more effective. A PSI-wide system for community target nominations has been instituted, and there are plans for PSI-wide coordinated promotions, such as for technologies and methods, at meetings.
Community Interactions. Subcommittee Chair John Moult discussed opportunities for better distribution of structures and technologies to the community, but noted the great improvement that has come with the KB onset. Additional opportunities are there to interact with community biologists, e.g. with respect to metagenomics and for outreach at national meetings.
Operations and Management Group (OMG). OMG has no chair, but Stephen Burley and Ian Wilson described OMG activities on behalf of the Subcommittee. This small committee meets frequently, often weekly, to coordinate target selection for the large-scale centers; and to address evaluations, such as the PSI Assessment Report; to prepare for future directions, as with the Futures White Paper; and to address other common concerns. An especially notable achievement of the past year was the leadership of OMG in establishment of the KB.
Bioinformatics Group (BIG). Like OMG, BIG has no chair, but Adam Godzik was present to represent the group. BIG holds weekly phone calls to coordinate target selection and to find agreement on definitions and computational procedures for metrics of coverage and leverage. BIG runs the drafts for large-scale center targets, recently focusing on very large families with few structural representatives and high biomedical interest (MEGA) and for metagenomics families (META). It is striking to note that these elite bioinformatics experts have subordinated their own competitive technical efforts to the common good of the PSI effort.
Knowledgebase (KB). The KB is overseen by the Structural Genomics Knowledgebase Steering Committee, which is not a Subcommittee of the PSI Steering Committee, but Ed Lattman who chairs the KB Steering Committee agreed to bring us up to date. From his perspective, the KB is off to a good start. Of course, it must provide the experimental information, but it offers added value as a marketplace for ideas. There is a need to preserve in any future plan, and, ultimately, the KB should cover and promote all of structural biology.
PSI Future Plans
All of the time available on the second day (8:30-11:45am) was devoted to a structured discussion of the future for PSI, including post-PSI-2 opportunities for NIGMS initiatives. Because of the time limitation and the breadth of factors to consider, the discussion was divided into specified topics with designated discussion leaders, and each topic had a strictly held time limitation. Leaders set the stage with a few comments, and active participation ensued on each topic from the assembled group.
Structural Coverage (John Moult, discussion leader). There remains within PSI a strong appreciation for the usefulness of structural coverage targets for the initiative, but there is also a great appreciation within PSI as in the broader biological community for the need to have stronger connection to biology in the results. Part of this seems to be a communication problem on the value of structures, particularly ones of unknown function, and KB is helping this. This might be done by having greater focus on human disease related proteins, for example, and for an increased focus on families found only in eukaryotes.
Modeling (Adam Godzik, discussion leader). There are differing perceptions on the quality and utility of models, but it is clear that attention needs to be paid to improvements in the methodology and in making tools for creating models readily accessible to the community. It is also noted that modeling methods can incorporate ‘low resolution’ experimental data to improve both. Modeling is equally important for future initiatives more closely focused on biology and function as it is for the utility of structure determined as representative in a broad coverage initiative.
Biological Topics (Ian Wilson, discussion leader). The first observation is that “everything is biological,” and the PSI focus on numbers as a driving force may detract from the biological content of the results. Biology comes from the coverage program as well as from efforts focused on themes of biology and functions. The efficiency of high-throughput structural efforts must somehow entail a structure leading biology rather than the reverse as in conventional structural biology. On the other hand, better communication is not the only answer; there does seem to be a need to do more in target selection to assure that results will have high functional coverage as well as high structural coverage.
Challenging Proteins (Wayne Hendrickson, discussion leader). There is a strong consensus that membrane proteins and protein-protein complexes remain as challenging subjects for structural analysis, but ones of such high value that they must be a component of any future initiative. Soluble eukaryotic proteins are also of high value, but it seems here that while challenges certainly do remain these problems can be integrated now into standard pipeline processes.
Technologies (Lance Stewart, discussion leader). There is a sense that technology development for structural genomics works best in the context of pipelines themselves. New technologies, perhaps outside the pipelines, will be needed for challenging problems such as protein-protein complexes; so there is a need for a mix of central technology development and more directed technology development. Consideration should be given to making the tools from the centers more useful for biologists broadly.
Project Characteristics (Andrzej Joachimiak, discussion leader). Center-level projects for this area need to have characteristics that take advantage of the HTP aspects of PSI pipelines. One-off projects would greatly reduce the efficiency of PSI centers. Structures come fast, but function is slower; thus, there are needs for thoughtful processes to connect the two using generic functional analyses in center settings.
International Perspectives (Wayne Hendrickson, discussion leader). Messages solicited from Committee members Stuart and Yokoyama were reported to give views from parallel effort in Europe and Japan, respectively, and the work of the Structural Genomics Consortium (SGC) was also discussed. Our lack of knowledge about what is happening now in China was also mentioned. Despite differences in approach, there is a need for PSI to take a leading role in international coordination, for example by making PepcDB become international, as Target DB already is, and by working toward coordination with SGC on human proteins if they become a focus of a new NIH initiative.
Community Involvement (Stephen Burley, discussion leader). There are examples of very fruitful community involvement piloted by current PSI centers, notably NYSGXRC with the amidase/hydrolase community and JCSG with the Thermotoga community, but there should be broader, program-wide efforts in this direction.
Organizational Models (Guy Montelione, discussion leader). This topic elicited a lively discussion with participation of many. Suggested models included from having separate specialized centers for biological theme projects, having large-scale production centers that propose biological themes of emphasis with auxiliary centers organized for biological collaborations, having options for biological projects/centers added in later years after production centers are defined. No consensus was achieved except that the highly cooperative spirit and network character of the enterprise should be preserved.
Submitted by Wayne A. Hendrickson (Chair) on behalf of the PSI Steering Committee