Clinical OMICS

MAY-JUN 2018

Healthcare magazine for research scientists, labs, pathologists, hospitals, cancer centers, physicians and biopharma companies providing news articles, expert interviews and videos about molecular diagnostics in precision medicine

Issue link: https://clinicalomics.epubxp.com/i/979295

Contents of this Issue

Navigation

Page 32 of 51

www.clinicalomics.com May/June 2018 Clinical OMICs 31 clinical impact of each mutation. To date, Mastermind has indexed nearly 6 million scientific articles covering every disease, gene and variant, out of the 30 million titles and abstracts in PubMed. The indexed articles contain data on more than 1.5 million variants, according to Genomenon. "Our partnership is starting with enhancing the Mastermind Genomic Search Engine to determine the pathogenicity of variants by ACMG criteria," Klein said. "The latest release of Mastermind helps users to identify clinically relevant literature that is applicable to ACMG classification to accelerate variant interpre- tation. This is the first step in reducing the single big- gest bottleneck in scaling the clinical use of whole genome sequencing." At the American Associ- ation for Cancer Research (AACR) Annual Meeting 2018 in April, Genome- non and Veritas presented findings from a pilot study assessing Mastermind. In "Evaluation of Genom- enon Mastermind for Gene- Level Literature Curation," Ryan Schmidt, M.D., Ph.D., a Molecular Genetic Pathology Fellow at the Laboratory for Molecular Medi- cine at Harvard Medical School and a resident-physician-clinical pathology at Brigham & Women's Hospital, and colleagues, used three search strate- gies to examine 10 genes frequently included in diagnostic testing for hypertrophic cardiomyopathy with a range of known gene-disease asso- ciation strengths (MYH7, MYBPC3, TPM1, TNNI3, TNNT2, ACTN2, CSRP3, TNNC1, NEXN, and VCL). The researchers carried out a PubMed search; a PubMed search with medical subject headings (MeSH)-based disease terms, representing a curated vocabu- lary that is shared between PubMed and Mastermind; and a search via Mas- termind. The searches yielded 1,910, 1,436, and 2,432 PubMed reference numbers, respectively. Mastermind increased the number of results by 69.4% over a matched PubMed/MeSH search, and 27.3% over a "real world" PubMed search. The study also found that 22% of PubMed results were not found by Mastermind—but the percentage dropped to 3.9%, and number of arti- cles rose to 4,892 following improve- ments made to Mastermind in response to these results. "GM [Genomenon Mastermind] improves literature curation sensitiv- ity due to its expanded search capabil- ities including examination of the full text and may improve on traditional literature search methods in certain situations," Schmidt and colleagues concluded. "Additional [development] to improve the specificity of GM search results is required to fur- ther eliminate 'off-target' results." The companies have iden- tified ensuring proper vari- ant classification efficiently, at scale, by identifying and prioritizing relevant litera- ture as one of two challenges that have held back the advancement of the genom- ics industry—and which their partnership is designed to address. The other chal- lenge, the companies assert, is developing an efficient way to alert and update vari- ant classifications as new knowledge arises. In the partnership's sec- ond stage, Veritas and Genomenon plan to collabo- rate on using AI and machine learning to accelerate genomic interpretation. Genomenon has developed a propri- etary GLP engine based on technology patented and licensed from the Univer- sity of Michigan. The company's GLP engine is designed to recognize and index all synonyms that authors could use for diseases, genes, and variants (continued on next page) wildpixel / Getty Images

Articles in this issue

Links on this page

Archives of this issue

view archives of Clinical OMICS - MAY-JUN 2018