Christopher Cassa, PhD, is a graduate of the Harvard-MIT Division of Health Sciences and Technology program in Bioinformatics and Integrative Genomics. Dr. Cassa is an Instructor at Harvard Medical School, a research affiliate at Boston Children’s Hospital, an associate geneticist at Brigham and Women’s Hospital and a research affiliate at the Broad Institute. His laboratory conducts predictive analytic research in two major application areas: the pathogenicity assessment of genomic variants and the synthesis of public health information from infrastructure and social media. These applications both draw on “big data” approaches to enable unprecedented extraction of information from existing data sources.
One of the most important problems in modern clinical and translational medicine is the analysis of whole genome sequence data. Healthy individuals carry hundreds of genetic variants that have been previously associated with disease, and there is a pressing need to distinguish between causal variants and those that are either incompletely penetrant or false positives. We computationally assess the clinical impact of these genetic variants using previously unleveraged population health datasets and machine learning approaches. We have improved existing techniques by identifying a patient’s most likely pathogenic and clinically significant variants, and have separately assessed the clinical and syntactic validity of previously described variant databases.
In public health and public safety infrastructure, we develop and conduct research using the Health and Homeland Alert Network (HHAN) with the Massachusetts Department of Public Health. This project identifies and synthesizes relevant information for public safety and health officials in a timelier manner than existing techniques. We extract relevant information from large datasets that contain temporal and geographic information, including social media and emergency 911 data. We then transform individuals who are near the scene of incidents into “sensors” to provide rapid insight during natural disasters, disease outbreaks and ongoing conflicts. We develop classification strategies and filtering approaches that group relevant incidents, and compare newly observed data against temporally adjusted keyword frequencies to identify unusual spikes in keyword use.