Precise estimation of residue relative solvent accessible area from Cα atom distance matrix using a deep learning method

Abstract

Motivation The solvent accessible surface is an essential structural property measure related to the protein structure and protein function. Relative solvent accessible area (RSA) is a standard measure to describe the degree of residue exposure in the protein surface or inside of protein. However, this computation will fail when the residues information is missing. Results In this paper, we proposed a novel method for estimation RSA using the Cα atom distance matrix with the deep learning method (EAGERER). The new method, EAGERER, achieves Pearson correlation coefficients of 0.921-0.928 on two independent test datasets. We empirically demonstrate that EAGERER can yield better Pearson correlation coefficients than existing RSA estimators, such as coordination number, half sphere exposure, and SphereCon. To the best of our knowledge, EAGERER represents the first method to estimate the solvent accessible area using limited information with a deep learning model. It could be useful to the protein structure and protein function prediction. Availability The method is free available at https://github.com/cliffgao/EAGERER. Supplementary information Supplementary data are available at Bioinformatics online.

I-DNAN6mA: Accurate Identification of DNA N6-Methyladenine Sites Using the Base-Pairing Map and Deep Learning

“…The network architecture B(i) includes initialization convolution layers, block a repeated N 1 times, block b repeated N 2 times, block c repeated N 3 times, an inception module, 44 and one self-attention layer. 45 The initialization convolution layer, which is used before blocks a, b, and c, transforms the input feature maps into a spatial vector with a larger signal channel. Each block (a, b, and c) consists of two dilated convolutional layers with kernel sizes of 3 × 3 and 3 × 3, respectively, and the dilation rate (dr) of 2.…”

“…The 16 × L × L matrix is entered into the network B­(i) to dig out discriminative information of the base-paring channels map. The network architecture B­(i) includes initialization convolution layers, block a repeated N 1 times, block b repeated N 2 times, block c repeated N 3 times, an inception module, and one self-attention layer . The initialization convolution layer, which is used before blocks a , b , and c , transforms the input feature maps into a spatial vector with a larger signal channel.…”

“…Each channel, represented by a square matrix of the same dimension as the length of the input DNA sequence, indicates the occurrence of one of the 16 possible base pairs between the input nucleotides . Benefitting from this novel feature representation, the first stage B­(i) (see Figure ) of our proposed DL-based model, which is an ensemble of deep hybrid networks of ResNet , coupled with the “inception” module and self-attention networks, can explicitly grab all short- and long-range interactions and all possible base pairings. Our model includes one additional stage B­(ii) to store the output of B­(i) and concatenate it with one-hot encoding as input features and uses BiLSTM capture knowledge of time series dependencies between nucleotides in the DNA sequence.…”

Freeprotmap: waiting-free prediction method for protein distance map

“…Chen et al [6] use molecular docking simulation and graph representations of proteins based on contact maps discover two candidate drugs. Other common research work includes protein solubility prediction [17], key site prediction [18,19], protein identification [20], and protein disorder region identification [21].…”