In metagenomics communities of microorganisms are studied by observing random fragments of their genomes. The potential of metagenomics have increased with the introduction of next generation DNA sequencing and it constitutes today an important measurement technique in ecology, ecotoxicology and medicine. Metagenomics is however limited by high levels of technical and biological noise, the low number of samples and the high dimensionality of the observed data. In this project we will develop novel statistical methods for analysis of gene count abundance in metagenomes. We will model the complex structure of metagenomics data and develop robust estimators for the biological variation between metagenomes. We will also develop new methods for normalization of metagenomes with varying taxonomic composition. The methods will be applied in three collaborative research projects where metagenomics is an essential measurement technique. The methods will also be implemented in a software package that will be freely available for the microbiology research community. The statistical methodology developed within this project will improve the performance of metagenomics and thus the study of uncultured microorganisms in general.
Professor vid Chalmers, Mathematical Sciences, Applied Mathematics and Statistics
Funding Chalmers participation during 2012–2015