Sparsity, Epistasis, and Models of Fitness Functions - Leveraging Interactions in Protein Function Prediction

Explore cutting-edge approaches to understanding and modeling protein fitness functions in this comprehensive conference talk from the Broad Institute's Models, Inference and Algorithms series. Delve into the power of the Walsh-Hadamard transform and Graph Fourier transforms for analyzing epistatic interactions between amino acids. Learn how leveraging the natural sparsity of fitness functions can optimize experimental design and improve predictive modeling. Discover the innovative Epistatic Net method for regularizing neural network models of fitness functions. Gain insights into viewing protein function prediction through the lens of signal recovery and the Fourier transform. Understand how these advanced techniques can be applied to tackle the challenges of predicting biological functions from amino acid sequences, with potential implications for fields such as statistical genetics and computational biology.

MIA: Amirali Aghazadeh, David Brookes: Sparsity, Epistasis, and Models of Fitness Functions