PDF《Association Information Network》

Fig. 1). Applications underwent a multistage review process similar to that conducted by the NIH, including an administrative review for adherence to the application format and submission requirements, a scientific peer NATURE GENETICS | VOLUME

39 | NUMBER

9 | SEPTEMBER

2007 1045 ?

200 7 Nature Publishing Group http://www.nature.com/naturegenetics C O M M E N TA RY review assessing the likelihood of identifying important genotype-phenotype associations from the study, and a programmatic and tech- nical review for quality and ease of use of the proposed data set, limitations on research use or data sharing, public health significance and diversity of represented populations. Applicants submitted an online application describing the study design and population, the disease proposed for study, the available phenotypic and environmental exposure data, the proposed analytic strategies and follow- up studies and the willingness of contributing investigators and their institutions to abide by GAIN policies.In addition,to verify data avail- ability and assess data quality, all applicants submitted electronic files of de-identified, individual-level participant data and accompa- nying documentation before their applications were referred for peer review. Informaticians at the National Center for Biotechnology Information (NCBI) assessed the completeness and accuracy of data trans- fer,the extent of missing and out-of-range val- ues, the quality of documentation, the extent of human curation needed to integrate the documentation with the phenotype data and the clarity of case and control definitions.This initial assessment was provided to peer review- ers and GAIN project staff, and NCBI person- nel were available as needed during the peer review discussions. Scientific peer review was conducted by an independent committee of leading scientific experts from government, industry and aca- demia. Applications were evaluated on several criteria, including significance and complex- ity of the trait, the need for a GWA study, the appropriateness of the study design and population(s),the quality and completeness of phenotype and exposure measures to be pro- vided to GAIN,the strength of the evidence for a genetic component for the trait, the antici- pated size of a genetic effect and power to detect it,the advantages in terms of strategies for data management and data analysis and the advan- tagesforreplicationstudiesandfollow-upstud- ies to identify the causative genetic variant(s). Applicantswereprovidedwithabrief summary of reviewers'

comments after the review. Based on initial peer review results, the GAIN Steering Committee identified a subset of applications to be considered for participa- tion in GAIN.These were referred to a technical advisory group (TAG) composed of technical expertsfromacademia,governmentandindus- try with expertise in genetics, epidemiology, bioethics, biostatistics and GWA studies. The TAGconductedanin-depthexaminationof the highest-priority applications,worked with the applicants to optimize the suitability of their Feb 8CMay 9: solicitation June 18C19: initial peer review July 11CAug 30: TAG review Sept 5COct 5: Steering Committee, FNIH board review Oct 12: announcement of initial studies OctCJan: Submission of samples JanCMar: genotype HapMap samples FebCNov: genotype GAIN samples June: data available F M A M J J A S O N D J F M A M J J A S O N D J F Phase I: Phase II: Phase III: solicitation review and selection genotyping Phase IV: dissemination and analysis Figure

1 Timeline for GAIN solicitation, genotyping and release of data from initial 18,000 samples. proposed design for GAIN and provided rec- ommendations to the Steering Committee to support the final selection process. Unlike the typical NIH peer review process,the TAG could recommend that FNIH negotiate inclusion of samples and data in addition to, or instead of, those proposed in the initial application as well as recommending other design modifica- tions to maximize the scientific potential of successful applications. The TAG examined four areas in depth: (i) subject ascertainment, (ii) DNA specimen quality, (iii) power and analysis and (iv) data sharing and consent.Sample ascertainment cri- teria included sources, representativeness and comparability of cases and controls;