Discovery of novel cis-regulatory RNA structures using deep neural network

We have developed a comparative genomics model, EvoFam, based on modelling the characteristic evolutionary mutation pattern of conserved RNA structures and performed the first genome-wide screen for families of human RNA structures. In this project, we are developing new machine learning models to discover conserved families of cis-regulatory RNA structures across vertebrate genomes, which involves developing new RNA structure modelling and comparative genomics methods based on powerful neural network architectures for genome-wide discovery of structural RNA families. This will substantially enhance the sensitivity of our existing methods for structural RNA family discovery, leading to a greatly expanded catalogue of structural RNA families.

This project is a collaboration with Brian Parker.