VI based evolutionary analysis of cancer and metastasis

The Bayesian approach with Monte Carlo Markov Chain (MCMC) based inference has been hugely successful for phylogenetic tree reconstruction.

Since cancer development is due to somatic evolution, phylogenetic tree reconstruction is becoming increasingly important in cancer research. Evolutionary analysis of cancer holds the potential to reveal crucial information regarding metastasis.

However, speed is essential to Bayesian phylogenetics, particularly for future massive single-cell data sets in medical investigations. Opportunely, the Variational Inference approach yields fast approximate Bayesian inference of high quality. It has often been successfully used to deliver impressive performance gains for Bayesian inference compared to MCMC.

Although it earlier was unknown how to apply VI for phylogenetics, recent advances, especially our fast VaiPhy approach, and the associated computational efficiency gains open up novel research opportunities.

We will unlock the potential of VI for rapid Bayesian analysis in phylogeny and integrated phylogenetic analysis, especially in cancer research.

The paramount clinical relevance of this method development is that it will pave the ground for an improved ability to predict whether a tumor has metastasized or not, based on its molecular characteristics.

We will supply the necessary computational tools to researchers approaching such questions or support them by active collaboration.