Protein model quality evaluation method based on deep learning

Abstract

A protein model quality evaluation method based on deep learning comprises the steps that firstly, a protein index with the set protein length, the maximum sequence redundancy and the resolution ratio is screened out from a PISCES server, and then corresponding protein structure information is downloaded from a PDB library; the method comprises the following steps: generating a bait structure of each protein on different model mass distributions by using three methods of carrying out comparative modeling by using Rosetta CM, carrying out disturbance by inserting a fragment into a random position of a natural structure, and carrying out deep learning guidance folding by trRosetta, and constructing a data set; extracting one-dimensional, two-dimensional and three-dimensional feature information of each protein natural structure and bait structure thereof in the data set; output tensors generated through a series of three-dimensional convolution layers are flattened, are connected with other one-dimensional features in series, then are subjected to vertical and horizontal striping, and are combined with other two-dimensional features to obtain a 141 * L * L feature map, and the 141 * L * L feature map is input into a two-dimensional convolution residual network for training. The method is high in prediction efficiency and accuracy.

Description

technical field [0001] The invention relates to the fields of bioinformatics and computer applications, in particular to a protein model quality assessment method based on deep learning. Background technique [0002] Proteins are ubiquitous in almost all biological processes. Determining their structure and function helps to understand and potentially control these processes. However, although the determination of protein sequences is now a routine procedure, it is often very difficult to use this information to extract relevant functional knowledge of the systems under study. In fact, the function of a protein depends on a combination of its chemical and mechanical properties, which are determined by its structure. Therefore, the identification of protein structures from their sequences is of great importance, albeit a difficult task. Experimental structure identification is not feasible in all cases and is often very tedious and expensive. Consequently, computational m...

AI Technical Summary

Problems solved by technology

[0005] In summary, the existing protein model quality assessment methods are not perfect in terms of the accuracy and computational efficiency of protein model quality assessment, resulting in the inability to reliably obtain protein model quality and then guide protein refinement, so improvements are needed

Images