Statistical and Computational Methods for Marine Ecosystems

Christian Müller
Helmholtz Zentrum & LMU
München, Flatiron Institute

A paramount challenge in modern oceanography is the analysis and integration of its large and diverse datasets, ranging from global remote sensing data to sequencing-derived microbial abundances from cruises or marine stations. Statistical modeling, paired with modern machine-learning and optimization methods, can play a pivotal role in combining these large-scale datasets to gain a holistic, coherent picture of the underlying marine ecosystems and their interactions with the environment. The CBIOMES collaboration provides an ideal environment for fostering such data-and model-driven research in marine ecosystems thanks to the high degree of multi-disciplinary expertise.

Over the first funding period of CBIOMES, the Müller group established strong collaborations with several CBIOMES groups and contributed statistical modeling, optimization, and software expertise to several (ongoing) projects, in particular those involving microbial sequencing data. In Phase II we look forward to building on these collaborative projects with a view to exploring unsupervised learning as a tool to identify ocean provinces from remote sensing data, building statistical models to explore ocean microbiome-environment associations, using tree-aggregated learning for causal inference in ocean data, and joint modeling of flow cytometry and amplicon data.

CBIOMES Collaborators working with Müller

News from the Müller Group