My research focuses on studying specific characteristics of biological data:
Poisson hurdle model-based clustering for microbiome data
(Submitted to Bioinformatics)
- We developed a Poisson hurdle model-based clustering method for count datasets to deal with its high sparsity.
Dirichlet Multinomial Biclustering for Compositional Microbiome Data
- To mimic the sampling structure of microbiome datasets, we introduce a model-based biclustering method based on a Dirichlet Multinomial dstribution.
Bayesian Feature Selection and Causal Inference for Group Structured Data
-
In high-dimensional regression with subgroups of observations, we try to identify differentially expressed features by assigning a joint Lasso penalty.
-
We use a fully Bayesian approach to rewrite the joint Lasso regression, do feature selection and construct confidence intervals for features of interest.