Biostatistics & Computational Biology Branch
Research Summary
Shyamal Peddada is a Senior Investigator who leads the Constrained Statistical Inference (CSI) Group within the Biostatistics and Computational Biology Branch. The group focuses on developing broadly applicable rigorous biostatistical methods that are inspired by biomedical and environmental health research. Methods developed by Peddada’s group have applications to toxicology, epidemiology, various omics data and others. In addition to methodological research, the group is engaged in various scientific collaborations in biomedical and environmental health research. A research area of particular interest is to understand the role of human microbiome in health and disease.
Methods developed in this group exploit the underlying constraints in the scientific question or data. Constraints arise naturally in many scientific investigations either due to the underlying study design and scientific hypotheses of interest, such as in a dose response study or in a time course experiment; or due to the intrinsic characteristics of variables under investigation, such as the expression of a gene participating in a cell-division cycle or in the circadian clock; or due to the underlying technology, such as the scRNA-seq, 16S or metagenomics microbiome data, and others. Statistical methods that exploit such constraints are substantially more powerful than routine unconstrained statistical methods such as the standard linear regression, ANOVA, logistic regression or standard non-parametric methods. Equivalently, the constrained statistical inference-based methods require substantially smaller sample size to achieve the same power as the standard methods. Hence, they potentially require fewer biospecimens and are cost effective. More importantly, in many instances these constrained inference-based methods provide better scientific interpretation of the data than the standard methods.
Peddada’s group develops parametric and non-parametric constrained inference-based methods in low as well as high dimensions and applies the resulting methodologies to a wide range of data. Some examples include gene expression studies in toxicology, microbiome studies related to infant gut, infectious diseases, chemical exposures, and others.
Several user-friendly and freely-downloadable software have been developed by Peddada’s group such as ANCOM, ANCOM-BC, ANCOM-BC2, SECOM, ORIOGEN, CLME, ORIOS, and others.
In addition to conducting methodological and collaborative research, as well as developing user-friendly software, the CSI group is actively engaged in mentoring trainees at all levels who are enjoying successful careers at various universities, research institutions and industries.