Protein classification method

Abstract

The invention discloses a protein classification method. The method comprises the following steps: putting a protein three-dimensional model into a voxel occupation grid with the size of N*N*N; projecting N*N rays to the voxel occupation grid, setting each ray to pass through the centers of N voxels; constructing an occupation model based on the intersection condition of the rays and the protein three-dimensional model; and classifying the proteins by the occupation model through a three-dimensional convolutional neural network. Volume representation of a protein structure is realized by adopting the voxel occupying grid, a rapid and accurate classifier is created for protein from original protein volume data by utilizing machine learning, and a high-speed and high-precision classificationeffect is realized for a protein data set.

Description

technical field [0001] The invention belongs to the technical field of protein data classification, and in particular relates to a protein classification method. Background technique [0002] The goals of structural biology include a comprehensive understanding of molecular shape and form supported by biological macromolecules, and extending this to understand how most biological processes are represented using different molecular structures. Among these macromolecules, proteins are key effectors involved in most processes, with dynamically complex surfaces; they can consist of tens of thousands of atoms, at the atomic scale due to local (residue side chains) or global (loop or domain) structural changes that greatly affect their global and local shapes. [0003] Since the structure of proteins is linked to their function, and disruption of their interactions can lead to disease states, characterizing their shape is important to help identify potential binders, such as othe...

AI Technical Summary

Problems solved by technology

However, the current challenges for CNNs-based approaches to protein data classification are mainly two-fold: 1) while extending the basic approach of CNNs to volumetric data is conceptually straightforward, it is unclear which architectures and data representations will yield good performance;

[0006] 2) The volume is difficult to calculate

Images