The ability to modify cells from one type to another is known to have significant potential in the fields of regenerative medicine and disease modelling. In most cases, this has involved an intermediate step of creating a pluripotent cell (induced pluripotent stem cell).
Transdifferentiation, the process of converting from one cell type to another without going through a pluripotent state, has distinct advantages due to the reduced degree of change required, the speed of the conversion and the fact that the cell never has to pass through the pluripotent state. However, development has been inhibited by the difficulty in identifying the precise transcription factors, from the large number available, that are required to effect the required change.
We combine high-throughput sequencing data with gene regulatory networks using our proprietary mogrify technology in order to reprogram between any two human cell types. We first identify the core regulatory networks (CRNs) that underpin each of the cell types. Our algorithm then compares these CRNs between a desired source and target cell type and identifies which factors are required to induce a cell conversion. We have then experimentally proven these predictions in a number of cell types, for instance between fibroblast and keratinocyte (figure) and keratinocyte to microvascular endothelial cell (figure).
We believe that the integration of cutting edge machine learning with world-class experimental protocols is the key to highly efficient and complete conversions which is borne out by our work to date. As well as independently predicting the small number of previously existing conversions, our approach has developed a large number of new conversions and, as we scale up, this will continue to accelerate. Our data-driven approach can identify the key regulators for any cell conversion, facilitating highly efficient and complete reprogramming.
“A predictive computational framework for direct reprogramming between human cell types” Rackham et al. 2016 Nature Genetics 48(3), 331.
Step 1 - Network Construction
Step 2 - Factor Identification
Step 3 - Gene Expression Check
Step 4 - Cell Morphology Check