Causal inference and information theory in biology
This project aims at analyzing the causal structure of genetic regulatory networks of stem cells of plants using novel causal inference techniques to be developed here. Known methods for causal inference from statistical data usually require a large number of samples. Our preliminary work shows that it is in principle possible to infer causal relations from sample size one if the variables are high-dimensional, since algorithmic information provides additional hints on causal directions. Recent advances in genomic methods have allowed the simultaneous quantification of all genes in an organism.
To identify the causal relation between individual transcripts, we will use inducible expression to analyze the effect of the homeodomain transcription factor WUSCHEL on the regulatory network of plant stem cell control. After appropriate clustering of the genes, we obtain a causal network between extremely high-dimensional variables, to which algorithmic infor- mation theory based methods can be applied. The inferred causal relation will then be tested by advanced experiments.