The Genomics and Bioinformatics Group (GBG) represents the Center for Cancer Research's most substantial enterprise in bioinformatics. That bioinformatic activity is well-integrated with the experimental activities of the group. In fact, all of the algorithmic and software developments have been motivated by needs of the group's experimental science. Over the last several years we've developed the """"""""Miner Suite"""""""" as a state-of-the-art, web-based, portable, principally-Java set of tools and databases focused on needs that we found were not being served by other bioinformatic developments. Our web site, http://discover.nci.nih.gov, has been featured in Science and, locally, in the NIH Catalyst. The following is a list of the Miner Suite modules, ending with four that went online in 2007: CellMiner, SpliceMiner, AffyProbeMiner, and LeFEminer: * MedMiner accelerates searching and organization of PubMed literature 5-10 fold (L. Tanabe, et al.). * MatchMiner translates among gene identifiers for lists of genes from microarrays and other high-throughput omic platforms (K. Bussey, et al.). * GoMiner leverages the Gene Ontology for lists of genes (e.g., from microarrays) (B. Zeeberg, et al.). * CIMminer produces Clustered Image Maps (CIMs) (i.e., """"""""clustered heat maps""""""""), the ever-present icon of 'Postgenomic' biology. We introduced CIMs in the early 1990's. * LeadScope/LeadMiner links molecular targets (e.g., in the NCI-60 cancer cell lines) to 27,000 chemical substructures in a fluently browsable format (with P. Blower, et al.). * MIMminer provides searchable electronic forms (eMIM's) of the elegant, scholarly Molecular Interaction Maps developed by K. Kohn, M. Aladjem, and Y. Pommier. * High-throughput GoMiner extends GoMiner's statistics and visualizations to encompass large sets of microarrays (e.g., from time course studies or clinical trials) (B. Zeeberg, et al.). * AbMiner provides a browsable relational database of available monoclonal antibodies that we have characterized. It includes our quality control results and multiple link-outs (S. Major, et al.). * SmudgeMiner diagnoses spatial artifacts on Affymetrix and spotted arrays (M. Reimers, et al.). * MethMiner provides pattern visualization and statistics for DNA methylation. Program not yet public. (S. Kim, et al.). Miner Suite programs that went online publicly in 2007 were as follows: * CellMiner provides the NCI-60 and other molecular profile databases in a SQL-searchable relational format, with metadata on the experimental platforms and cell types (U. Shankavaram, S. Varma, et al. -- see 2007 publication in Molecular Cancer Therapeutics). * SpliceMiner provides a robust infrastructure for dealing with transcript splice variation (A. Kahn, M. Ryan, et al. -- see 2007 publication in BMC Bioinformatics). * AffyProbeMiner provides what we believe to be the best tool for re-mapping Affymetrix probes to achieve sharper results from the commercially available Affymetrix arrays. It also integrates the re-mapped array data with GoMiner and High-Throughput GoMiner (H.-F. Liu, B. Zeeberg, et al. -- see 2007 publication in Bioinformatics). * LeFEminer is a web-based implementation of the LeFE (Learner of Functional Enrichment) algorithm, which uses the random forest machine-learning paradigm in a novel way to predict functional relationships from microarray and other high-throughput data types. (G. Eichler, et al. -- see 2007 publication in Genome Biology). Those programs, largely designed and implemented under a competed contract with SRA International, (contract #263-01-D-0050, CIO-SP2 Delivery Order 2313028) are freely available and used by thousands of investigators worldwide. The success of the programs is attributable in part to our adoption of the Agile software development paradigm, which promotes close, iterative interaction between software engineers, biologists, and bioinformaticists. That success is also partially attributable to adoption of Unit and System Testing methods; whenever code is re-deposited in our version control system, it's automatically subjected to more than 1700 tests to minimize the chance that changes made have broken something elsewhere in the overall code base. The development team won an SRA Project Excellence Award for the Miner Suite (1 of 4 awarded out of more than 700 competing projects). We're integrating the Miner Suite applications with a variety of public bioinformatic software projects, including caBIG, The Cancer Genome Atlas, and the CGEMS genome-wide association project. * caBIG: We were awarded two caBIG pilot grants (one to caBIG-enable GoMiner, the other to caBIG-enable our NCI-60 databases). The group, particularly lead software engineer David Kane, has made strong contributions to caBIG's Integrative Cancer Research and Architecture Working Groups. * The Cancer Genome Atlas: Working with TCGA development staff, we used SpliceMiner as the basis for development of bioinformatic infrastructure to support the integration of genotypic/phenotypic data of multiple types. That infrastructure has now been adopted by TCGA's Data Integration Committee for the purpose. * CGEMS: Working with bioinformaticists in the NCI Division of Cancer Epidemiology and Genetics (DCEG), we have developed a program package provisionally named ChromMiner for visual integration of SNP data from the genome-wide association studies with phenotypic data on the cancers (including gene expression and comparative genomic hybridization data). That program package is central to a proposal, which I was instrumental in drafting, to combine the epidemiological/genotypic expertise of DCEG with the phenotypic expertise of CCR. The proposal is titled """"""""DCEG/CCR Plan for Follow-up of Genomic Regions of Association Identified by CGEMS.""""""""
