# A deep learning approach to the structural analysis of proteins

## TL;DR

This paper introduces a deep learning model that predicts global properties of proteins, such as eigenvalues of fluctuation modes, aiding in structural analysis and identification of mechanically significant regions.

## Contribution

The study develops a neural network capable of predicting global structural properties of proteins directly from atomic data, addressing limitations of local descriptors.

## Key findings

- Successfully predicts eigenvalues of protein fluctuation modes
- Identifies mechanically relevant regions in proteins
- Demonstrates potential for global property analysis

## Abstract

Deep Learning (DL) algorithms hold great promise for applications in the field of computational biophysics. In fact, the vast amount of available molecular structures, as well as their notable complexity, constitutes an ideal context in which DL-based approaches can be profitably employed. To express the full potential of these techniques, though, it is a prerequisite to express the information contained in the molecule's atomic positions and distances in a set of input quantities that the network can process. Many of the molecular descriptors devised insofar are effective and manageable for relatively small structures, but become complex and cumbersome for larger ones. Furthermore, most of them are defined locally, a feature that could represent a limit for those applications where global properties are of interest. Here, we build a deep learning architecture capable of predicting non-trivial and intrinsically global quantities, that is, the eigenvalues of a protein's lowest-energy fluctuation modes. This application represents a first, relatively simple test bed for the development of a neural network approach to the quantitative analysis of protein structures, and demonstrates unexpected use in the identification of mechanically relevant regions of the molecule.