Constrained Statistical Inference with Applications

Eunice Kennedy Shriver National Institute Of Child Health & Human Development

Investigators

Abstract

In many applications, researchers conduct multi-group studies where the experimental groups can be either nominal or can be ordered such as by disease severity, dose, time, etc. In all such cases, researchers are interested in performing multiple pairwise comparisons of high dimensional outcome variables between groups. In the case when the experimental groups are ordered, such as disease severity, dose, time, etc., researchers are interested in identifying patterns in the outcome variables over the ordered groups. In this project we are making progress in developing methods that address the following specific questions: (a) Methods for differential abundance analysis of microbiome data for multi-group studies. (b) Methods for describing associations among microbes within and between experimental groups. (c) Methods for analyzing single-guided RNA data obtained from CRISPRi technology. (d) Methods for describing nonlinear shapes in the relationships between placental hormones during pregnancy and various infant outcomes.