Bayes Regularization and Microbiome Multi-Omics: A Two-Tailed Narrative - Dr. Mallick

Biostatistics Branch Seminar Series

April 21, 2020 | 9:00 AM – 10:00 AM

NCI Shady Grove Webex only

Add to Outlook Calendar

Speaker

Himel Mallick, Ph.D.,
Senior Scientist, Biostatistics,
Merck Research Laboratories

Abstract

In this talk, I will present a statistical narrative of two seemingly different but equally exciting topics: Bayes regularization and microbiome multi-omics, drawing upon my PhD and postdoc experience of researching in two vastly different academic settings. In the first part of the talk, I will revisit the first principles of variable selection and discuss regularization from a Bayesian vantage point. Several case studies involving both real and synthetic data will be presented for illustration purposes.

In the second part of the talk, I will take a quick detour to discuss the challenges of microbiome multi-omics, which are typically noisy, sparse (zero-inflated), high-dimensional, and extremely non-normal, often arising in the form of count or compositional measurements. I will specifically discuss two recent statistical approaches to data integration in human microbiome studies with an application to the multi-platform genomics data from the recently completed NIH Human Microbiome Project (iHMP).

Finally, I will conclude with comments on the promises and implications of scalable Bayes for large-scale multi-omics data integration and inference for translational epidemiology studies and provide some empirical evidence of using multi-omics both as a multi-purpose biomarker and potential therapeutic target in precision medicine.

Join Via WebEx
Meeting number: 478 574 156
Password: BBsem04.21

Join by phone
1-650-479-3207 Call-in toll number (US/Canada)
Access code: 478 574 156