Justin Silverman

MD-PhD Candidate

Duke University

Biography

Justin Silverman is a MD-PhD candidate at Duke University. His PhD work is being done in the Program of Computational Biology and Bioinformatics under the mentorship of Dr. Lawrence David and Dr. Sayan Mukherjee. Justin’s dissertation work focuses on longitudinal modeling and experimental design of host-associated microbiota.

In his free time Justin enjoys exploring the outdoors (backpacking, canoing, rock climbing), creating puzzles, or simply relaxing by Jordan Lake with his wife (Rachel) and their dog (Gauss). Justin and Rachel also maintain a blog statsathome.com.

Interests

Bayesian Statistics
Probability Theory
Time Series Analysis
Compositional Data Analysis

Education

MD, 2020

Duke University
PhD in Computational Biology and Bioinformatics, 2019

Duke University
BS in Physics and Biophysics, 2011

Johns Hopkins University

Selected Publications

Statistical Considerations in the Design and Analysis of Longitudinal Microbiome Studies

In this review we takle statistical considerations starting from the data generation process, discussing technical and biological variation, experimental design, and various modeling approaches currently available.

Justin D Silverman, Liat Shenhav, Eran A Halperin, Sayan Mukherjee, Lawrence A David

On bioRxiv

Details PDF

Dynamic linear models guide design and analysis of microbiota studies within artificial human guts

Here we introduce Multinomial Logistic-Normal Dynamic Linear Models (MALLARDs) for the analysis of longitudinal microbiome studies. We apply this framework for the analysis of a longitudinal study of 4 artificial gut models.

Justin D Silverman, Heather Durand, Rachael J Bloom, Sayan Mukherjee, Lawrence A David

On bioRxiv

Details PDF Code Dataset

A phylogenetic transform enhances analysis of compositional microbiota data (PhILR)

Surveys of microbial communities (microbiota), typically measured as relative abundance of species, have illustrated the importance of these communities in human health and disease. Yet, statistical artifacts commonly plague the analysis of relative abundance data. Here, we introduce the PhILR transform, which incorporates microbial evolutionary models with the isometric log-ratio transform to allow off-the-shelf statistical tools to be safely applied to microbiota surveys.

Justin D. Silverman, Alex D Washburne, Sayan Mukherjee, Lawrence A David

In eLife

Details PDF Code Figure Supplements

Recent Publications

Statistical Considerations in the Design and Analysis of Longitudinal Microbiome Studies
Details PDF
Dynamic linear models guide design and analysis of microbiota studies within artificial human guts
Details PDF Code Dataset
The Bayesian Sorting Hat: A Decision-Theoretic Approach to Size-Constrained Clustering
Details PDF Code Dataset Supplement
Phylogenetic factorization of compositional data yields lineage-level associations in microbiome datasets
Details PDF Code
A phylogenetic transform enhances analysis of compositional microbiota data (PhILR)
Details PDF Code Figure Supplements

Recent & Upcoming Talks

Statistical and Geometric Methods for the Analysis of Sequence Count Data
Oct 10, 2018, Fall Seminar Series, CCMB at Brown University
Statistical and Geometric Methods for the Analysis of Sequence Count Data
Aug 23, 2018, Seminar Series in the Center of Statistical Genetics in the Department of Biostatistics at the University of Michigan
Quantifying and Controlling for Sources of Technical Variation and Bias in Longitudinal Microbiome Surveys
Jul 29, 2018, JSM 2018
Statistical and Geometric Methods for the Analysis of Sequence Count Data
Apr 4, 2018, University of Pennsylvania Statistics Seminar Series
Geometric Methods for Modeling Time Evolution in Human Microbiota
Mar 25, 2018, ENAR 2018 Spring Meeting
Time-Series Models for Microbiome Data
Jun 9, 2017, CoDAWork 2017

Contact

Justin.Silverman@duke.edu