Open Lecture — Decoding the Human Microbiome

Our bodies harbor 100X more microbial genes than human genes, and this “microbiome” is integral to human biology. High-throughput sequencing provides a means to study our resident microbes. The data from each metagenomics experiment is a pool of millions of short (~100 base pair) DNA sequences sampled from the genomes of >1000 different types of microbes, without information linking each sequence to its source organism. Deconvoluting this mixture of data into estimates of the abundances of different microbes and microbial genes is a challenging problem, especially if estimates are to be quantitatively compared across samples. I will discuss several sources of bias in commonly employed estimators and some recently proposed solutions. These methods reveal cryptic variation in the microbiome, some of which is correlated with disease.

Light refreshments will be served before the lecture at 3:30 p.m.