Aakarsh Vermani Rose Hills
Mutation Rate Prediction with Deep Learning
The rate at which DNA mutations occur is highly variable at different positions of the human genome. An accurate characterization of localized mutation rates could enable us to infer evolutionary histories, identify disease-associated genes, and even predict viral evolution, helping prevent pandemics like COVID-19.
The heterogeneity of rates across the genome and the myriad of factors affecting them complicate this task, but the recent success of biological language models on related tasks like variant effect prediction signifies they could be a valuable tool for predicting mutation rates.
Our preliminary results in the Song Lab suggest a deep learning model could outperform existing mutation rate prediction methods while being more applicable to non-human species. Not only could this lead to advancements in the applications mentioned earlier, but it would provide valuable insight into the mechanisms of DNA mutation and the interplay between neutral mutation rates and natural selection.
Message To Sponsor
Thank you so much for your generous donation! I’m incredibly grateful and excited to pursue this research opportunity this summer. I’ve wanted to be a scientist for as long as I can remember, and your contributions are making that dream come true.