Natural Sciences
Life Sciences
Scientific Computing
Scientific Computing

Prof. Susan Holmes, Department of Statistics, School of Humanities & Sciences, Stanford University, Stanford, California, USA

EMBL, Large Operon, Meyerhofstraße 1

Wolfgang Huber

We overcome the heterogeneity of the sources of information (NGS, trees, graphs, histories, metabolomics) using iterative structuration and sparse projection methods. Our models enable effective denoising of NGS data and the removal of contaminants. Tree-aware projections enable the combination of phylogenetic information and temporal and spatial dependencies. I will show examples of how our approach has enabled us to perform high resolution analyses of longitudinal data, and in particular, how we were able to discover biomarkers for preterm birth. These results are medically relevant and were recently confirmed in a completely independent replications study. The research focus of the Holmes group at the Department of Statistics in Stanford is realistic analyses of heterogeneous data and quantification of uncertainty for complex biological data. Our software and data are all open-source and can be downloaded and analysed using the R and D3 tools we have developed (dada2, phyloseq, treelapse).

Event data:
Import event data into Outlook Calendar