Protein Structure Initiative Advisory Committee (PSIAC): Recommendations for the Future of the PSI

January 2009

The PSIAC has followed the evolution of PSI-1 and PSI-2 closely and observed the impressive development of a new type of science: large teams working in a cooperative fashion, with clearly articulated goals, and with close management by NIGMS staff. The capacity of the resulting large PSI centers to carry out high-throughput structure determination is truly stunning, and the accomplishments from eight years of effort are outstanding in terms of numbers of new structures. Particularly impressive are the number of novel folds and the resulting impact on fold space, as well as the technological enhancements that enable high-throughput structure determination and consequently touch the entire field of structural biology. Moreover, the existing large-scale centers established as a result of PSI-1 and PSI-2 represent a new resource, unknown before the PSI, and one that changes what our scientific enterprise is capable of from 2009 forward.

In addition to the achievements of the large centers, considerable progress has been made by the several specialized centers, although the global impact is not so visible because each has less quantitative and less cooperative goals, and two are dedicated to very challenging targets, namely membrane proteins. Nonetheless, taken together these centers have also had a significant impact on the field.

The scientific community in large measure acknowledges and applauds the impact of PSI-1 and -2, but both formally, via the Assessment process (culminating in a report last January), and informally, via published commentaries, the Futures meeting, and direct communication, clearly has objections to continuing the PSI program in its current form. The best way to describe these is as a cost-benefit analysis. The impressive advances of PSI-1 and -2 are a direct result of substantial directed investment by NIGMS. Yet investment in this major program represents a choice, and the expenditure of funds to support this coordinated effort puts tension on the NIGMS portfolio during times of tight funding. The PSIAC emphasizes that decisions about the future of PSI required careful critical review of the successes of PSI and consequent appraisal of the potential impact of any future structural genomics ventures. The PSIAC has been very impressed by the extensive analysis of ongoing PSI programs over the past year and by the efforts of NIGMS staff to stimulate discussion and solicit input about structural genomics prospects and the best ways NIGMS may foster these. Now, the PSIAC emphasizes that impending decisions about any continued NIGMS investment in PSI must acknowledge and address the issues raised during this wide-ranging analysis of both PSI outcomes to date and future opportunities.

An overriding concern emerging from all consideration of PSI is the connection of results to biological questions. The PSIAC agrees with the Assessment team that the emphasis in the original RFAs for PSI-1 and -2 on annual goals, which included numbers of structures and cost per structure, is inappropriate for any future PSI RFA. Instead, any "PSI-3" RFAs must require a tight linkage of a biomedical question with target selection and structural determination goals. The questions to be addressed should be selected so that high-throughput structure determination is strategically and scientifically advantageous. A number of such questions have been identified in the White Paper prepared by the current PSI center directors, and others were discussed at the Futures meeting. The PSIAC refers the Council to the White paper "Opportunities for Structural Genomics Beyond 2010: Creating Partnerships for the Future" and to the summary of the "Futures" meeting prepared by Peter Preusch. Still more questions can be expected to emerge from the scientific community more broadly in an ongoing way.

The biological motivations for structural genomics may include the aim to achieve structural coverage, if the impact of the structural coverage on functions is well articulated and the relationship of new structures to available data, for example on a given model organism or a pathway, is well argued. Homology modeling methods are intimately tied to the use of structural data, and the PSIAC concludes that any RFA for "PSI-3" centers should stipulate that applicants provide a plan for the use of structures determined, and provide or develop methodology to enable a greater impact through homology modeling.

A second major concern emerging from the analyses of PSI-1 and -2 is the potential for inappropriate and potentially unfair conflict between groups investigating a given biological problem via the more traditional R01 or P01 funding mechanisms and those working on the same problem in a large team supported by PSI-type cooperative agreements. Ironically, this latter concern is exacerbated by the likely greater emphasis in a future PSI program on biological questions. There is widespread support in the community for the tenet that biological questions to be tackled by PSI-3 teams should be subject to full peer review. The PSIAC concurs that it is essential that selection of biological questions to be funded in a PSI-3 program be based on a full peer review process. Moreover, the PSIAC feels that it is crucial that anyone in the community be encouraged to submit proposals of biological projects for which competitive review will be based on scientific significance combined with the fit to a structural genomics approach. Such a gathering of scientific problems from the scientific community widely will greatly leverage the outcomes of structural genomics investment by NIGMS.

The PSIAC deliberated extensively with NIGMS staff about how a "PSI-3" funding mechanism could continue to support large teams capable of optimizing and applying high-throughput structural genomics methods, and at the same time enhance the biological impact of the NIGMS structural genomics investment and hold the biological goals to the standards of peer review used for investigator-initiated research. These discussions drove home the advantages of the cooperative agreements for the management and coordination of the large centers in PSI-1 and -2. Consequently, it was concluded after considerable discussion that separate RFAs would be needed to accomplish the desired goals. One would be for the support of large centers under cooperative agreements, with improvements in technology and expectation of high-throughput structure determination as key components. These centers would be expected to dedicate some fraction of their effort (ca. 15% might be reasonable) to biological questions of each center's choosing, and some fraction to technological advances (ca. 15%), with the remainder of the time and effort (ca. 70%) going to collaborations on problems selected via a second RFA. These latter projects could be initiated by members of the center or by any investigators in the broader community. After some discussion, the PSIAC concluded that the projects funded under the second RFA would not be associated ahead of time with a particular center, but instead would be assigned to a center in a mechanism not unlike the current target selection and assignment process. The timing of these two RFAs was discussed at length, and there were viewed to be some advantages to simultaneous announcements and review. The alternative, first announcing and carrying out review of large centers and next announcing and reviewing biological collaborative structural genomics proposals, could also be envisioned to work.

The PSIAC does not foresee a need to solicit and fund specialized centers (including those for modeling) via a set-aside under the proposed rubric. With one very important exception, the goals of the current specialized centers are expected to be incorporated into large center proposals or be independently packaged in R01 proposals or as parts of applications based on biological questions. The exception is the support for membrane protein structure determination.

The overwhelming biological significance of membrane proteins and the huge lag in the number of their structures solved relative to soluble proteins led to extensive deliberations on how to use NIGMS funds to foster new advances both in high through-put method developments for membrane proteins and in determination of key membrane protein structures, which themselves would provide great insight into important biological questions. The experiences of the two specialized centers dedicated to membrane proteins convinced the PSIAC that increased investment is needed to make substantial progress in overcoming obstacles, and that breakthroughs will come from approaches that are quite distinct from the highly successful high-throughput strategies used on soluble proteins. In particular, it appears crucial to success with membrane proteins to tightly couple intimate knowledge of biological function with efforts to determine structure. For the same reason we suggest that each membrane center focus on specific biological systems and questions, which will enhance their likelihood of success in bringing membrane protein targets to successful structural determination. As a result of the unique needs for major steps forward with membrane proteins, and the great urgency for progress with these challenging targets, the PSIAC supports the issuing of a separate RFA for centers whose primary responsibility is membrane proteins.

The PSIAC feels that continued support for the relatively recently launched Knowledgebase and Materials Repository should be a component of any future PSI. These initiatives are key to the leveraging of results of PSI efforts and to the impact of PSI results on a wide community of researchers. There is a need to regularize the support for the Knowledgebase via an openly competed RFA.

Lastly, the PSIAC urges NIGMS to leverage its support for biologically motivated structural genomics research via collaborations, partnerships, and co-funding arrangements with other institutes of NIH as well as other funding agencies. The power of the structural genomics centers established under PSI-1 and -2 and the anticipated capabilities fostered by PSI-3 should be of great advantage to more disease-oriented institutes, leading to mutually beneficial partnerships.

Submitted by Protein Structure Initiative Advisory Committee member Lila Gierasch