Giovanni Parmigiani



Giovanni Parmigiani is the Chair of the Department of Biostatistics and Computational Biology at Dana Farber Cancer Institute. Professor Parmigiani is a statistician whose work creates statistical tools for understanding cancer data, with particular focus on genetic epidemiology and genomics. For example, he uses Bayesian modeling and machine learning concepts for predicting who is at risk of carrying genetic variants that confer susceptibility to cancer. For another example, he is interested in addressing the challenges of cross-study replication of predictions by constructing predictors that learn replicability from multiple studies. His overarching goals are to increase the rigor end efficiency with which we leverage the vast and complex information generated in today's cancer research; and to foster the use of data sciences as a common thread to facilitate interactions between fields and academic cultures.

Scientific Interests:

Models and software for predicting who is at risk of carrying genetic variants that confer susceptibility to cancer. Application to breast, ovarian, colorectal, pancreatic and skin cancer.

Statistical methods for the analysis of high throughput genomic data: analysis of cancer genome sequencing projects; integration of genomic information across technologies; cross-study validation of genomics results.

Statistical methods for comparative effectiveness research: comprehensive models for lifetime history of chronic disease outcomes; Bayesian meta-analysis; Bayesian causal inference; decision analysis.

Bayesian modeling and computation: multilevel models; decision theoretic approaches to inference; sequential experimental design and their application to adaptive and multistage studies in clinical and epidemiological research.

Scientific Working Groups:

Genomic IntegrationA group of faculty and students working on methods for integrating information from multiple studies, techniques, and aspects of the genome and transcriptome. Areas of interest have included cross-study reproducibility, via integrative correlation and integrative association; formal Bayesian meta-analysis via hierarchical models; and integration of diverse genomic data via gene set analysis. A great source of inspiration for these activities has come from the collaboration with the Kinzler/Velculescu/Vogelstein lab, who carried out the first genome-wide investigation of somatic genetic alterations in cancer, using an array of complementary technologies. Our software tools include mixture modeling (POE), meta-analysis (XDE) and somatic mutation analysis (CancerMutationAnalysis).

BayesMendel Group A group of faculty and students working on Mendelian models for cancer genes. It grew out of the work that Don Berry and I did on the BRCAPRO model when we were at Duke University. After BRCAPRO, the group developed models for colorectal cancer (MMRPRO) pancreatic cancer (PancPro) and melanoma (MELAPRO), led extensive validation projects for these models, and developed statistical methodologies targeted to this area. The group maintains and improves BayesMendel, a free and open-source environment for Mendelian risk prediction modeling. Current collaborations include the National Familial Pancreatic Cancer Registry at JHU, the Cancer Genetics Network, as well as Drs Euhus (UTSW), Hughes (MGH), Klein (JHU) and Anton-Culver (UCI).