Mimetic Neural Networks: A unified framework for Protein Design and Folding

TL;DR

This paper introduces MimNet, a reversible graph neural network architecture that simultaneously addresses protein structure prediction and design, improving results by leveraging recent folding models.

Contribution

The paper presents MimNet, a unified, reversible neural network framework that jointly solves protein folding and design problems, enhancing design accuracy with better structure estimates.

Findings

Improved protein design results on ProteinNet dataset

Reversible architecture enables joint structure and design optimization

Enhancement over previous state-of-the-art methods

Abstract

Recent advancements in machine learning techniques for protein folding motivate better results in its inverse problem -- protein design. In this work we introduce a new graph mimetic neural network, MimNet, and show that it is possible to build a reversible architecture that solves the structure and design problems in tandem, allowing to improve protein design when the structure is better estimated. We use the ProteinNet data set and show that the state of the art results in protein design can be improved, given recent architectures for protein folding.