Curating Genetic Association Literature for Common Diseases

摘要

Papers describing genetic associations with common diseases are currently being published at a rapid rate. These new papers add to an already large body of literature which includes candidate gene studies, genome wide association studies, review papers, and meta-analyses. Related papers describe the basic epidemiology of these common diseases, as well as interactions between genes and environment, other genes, and drugs (pharmacogenomics), all of which may affect disease predisposition and management. As the field of personalized genomics continues to grow and mature, this body of literature is being synthesized in various ways. One aspect of utmost importance is to track gene-disease associations over time to see if they are replicated in different populations by different authors. This has historically^1^ been a limitation of genetic association studies and if not addressed, can be a major barrier to the adoption of personalized genomics. Another key need is to systematically collect data on the magnitude of the gene-disease effect (typically an odds ratio), variant identifier, allele frequency, risk and non-risk alleles, and other key information from the papers. This is often made more complicated the way that the data is presented by authors (in particular, it is often difficult to tell which is the risk allele), as well as by DNA strand issues. We have built a literature curation database that addresses these two key needs, using Ruby on Rails with MySQL. Lack of consistent standards for the reporting of gene-disease associations, either by journal editors or other consortia or agencies, make automated computer curation infeasible at this time. Thus, Navigenics does all curation manually, employing a team of epidemiologists and human geneticists. To minimize human error, we have incorporated quality control measures, including two independent literature reviews, into the data collection system. Methods for the collection, interpretation, storage, and retrieval of genetic association data from large numbers of papers will be discussed.