Project information
A machine intelligence approach to mapping cell differentiation with multi-omics data

Logic-based modeling (e.g. Boolean networks) is a framework through which computational biology studies complex emergent biochemical processes. Traditionally, such models were constructed by hand with the help of domain experts. However, recent broad availability of detailed single-cell observation data makes research into automated inference of logic-based models extremely relevant. Here, we address the shortcomings of this emerging field while proposing a clear biological application in cell differentiation studies. Specifically, we challenge the 50 years of attractor-centric research in logic-based models by proposing a new formal characterization connecting differentiated and undifferentiated cell types. We then tie it to a novel inference workflow, which addresses the problem of identifiability (the ability to select a single ideal model) and co-development of gene regulatory networks together with the logic-based models. This project has great potential to revolutionize developmental biology and extract previously unrealized value from existing genomic datasets.