Professor Holmes is specialized in the use and development of multivariate statistical tools based on particular metrics and topologies for the analyses of heterogeneous biological data. A fellow of the Institute of Mathematical Statistics and the Breiman lecturer at NIPS, 2016 she has published more than 150 papers and has been a leader in the use of simulations and nonparametric statistical methods for complex biological data for over twenty years.
She collaborates with Professor David Relman to analyze data from longitudinal perturbation studies of the human microbiome. She has developed methods for making inferences from DNA sequences, cytometry data, images, gene expression patterns and bacterial abudances in conjuction with spatial, network, tree or temporal information. Her main focus has been the development of statistical and computational tools (phyloseq and dada2) tat integrate and denoise mutiple sources of information on the same samples for microbiome studies.
Her lab has had success in understanding the relations between preterm birth and strain abundances in the microbiome as well as the interaction between genetic, clinical and dietary factors through the analysis of metabolomic and taxa abundances. Strongly committed to the principles involved in the practice of reproducible research, Professor Holmes uses provenance tracking and R markdown in all her publications. She has recently finished a book with Wolfgang Huber: Modern Statistics for Modern Biology. This book, published by Cambridge University press provides methods for learning how to use R and Bioconductor for multivariate statistics in immunology, microbial ecology, infectious diseases and molecular biology.