IGPRED: Combination of convolutional neural and graph convolutional networks for protein secondary structure prediction.

There is a close relationship between the tertiary structure and the function of a protein. One of the important steps to determine the tertiary structure is protein secondary structure prediction (PSSP). For this reason, predicting secondary structure with higher accuracy will give valuable information about the tertiary structure. Recently, deep learning techniques have obtained promising improvements in several machine learning applications including PSSP. In this article, a novel deep learning model, based on convolutional neural network and graph convolutional network is proposed. PSIBLAST PSSM, HHMAKE PSSM, physico-chemical properties of amino acids are combined with structural profiles to generate a rich feature set. Furthermore, the hyper-parameters of the proposed network are optimized using Bayesian optimization. The proposed model IGPRED obtained 89.19%, 86.34%, 87.87%, 85.76%, and 86.54% Q3 accuracies for CullPDB, EVAset, CASP10, CASP11, and CASP12 datasets, respectively.

Protein ß-sheet prediction using an efficient dynamic programming algorithm.

Predicting the ß-sheet structure of a protein is one of the most important intermediate steps towards the identification of its tertiary structure. However, it is regarded as the primary bottleneck due to the presence of non-local interactions between several discontinuous regions in ß-sheets. To achieve reliable long-range interactions, a promising approach is to enumerate and rank all ß-sheet conformations for a given protein and find the one with the highest score. The problem with this solution is that the search space of the problem grows exponentially with respect to the number of ß-strands. Additionally, brute-force calculation in this conformational space leads to dealing with a combinatorial explosion problem with intractable computational complexity. The main contribution of this paper is to generate and search the space of the problem efficiently to reduce the time complexity of the problem. To achieve this, two tree structures, called sheet-tree and grouping-tree, are proposed. They model the search space by breaking it into sub-problems. Then, an advanced dynamic programming is proposed that stores the intermediate results, avoids repetitive calculation by repeatedly uses them efficiently in successive steps and reduces the space of the problem by removing those intermediate results that will no longer be required in later steps. As a consequence, the following contributions have been made. Firstly, more accurate ß-sheet structures are found by searching all possible conformations, and secondly, the time complexity of the problem is reduced by searching the space of the problem efficiently which makes the proposed method applicable to predict ß-sheet structures with high number of ß-strands. Experimental results on the BetaSheet916 dataset showed significant improvements of the proposed method in both execution time and the prediction accuracy in comparison with the state-of-the-art ß-sheet structure prediction methods Moreover, we investigate the effect of different contact map predictors on the performance of the proposed method using BetaSheet1452 dataset. The source code is available at http://www.conceptsgate.com/BetaTop.rar.

Efficient dynamic programming algorithm with prior knowledge for protein ß-strand alignment.

One of the main tasks towards the prediction of protein ß-sheet structure is to predict the native alignment of ß-strands. The alignment of two ß-strands defines similar regions that may reflect functional, structural, or evolutionary relationships between them. Therefore, any improvement in ß-strands alignment not only reduces the computational search space but also improves ß-sheet structure prediction accuracy. To define the alignment scores, previous studies utilized predicted residue-residue contacts (contact maps). However, there are two serious problems using them. First, the precision of contact map prediction techniques, especially for long-range contacts (i.e., ß-residues), is still not satisfactory. Second, the residue-residue contact predictors usually utilize general properties of amino acids and disregard the structural features of ß-residues. In this paper, we consider ß-structure information, which is estimated from protein ß-sheet data sets, as alignment scores. However, the predicted contact maps are used as a prior knowledge about residues. They are used for strengthening or weakening the alignment scores in our algorithm. Thus, we can utilize both ß-residues and ß-structure information in alignment of ß-strands. The structure of dynamic programming of the alignment algorithm is changed in order to work with our prior knowledge. Moreover, the Four Russians method is applied to the proposed alignment algorithm in order to reduce the time complexity of the problem. For evaluating the proposed method, we applied it to the state-of-the-art ß-sheet structure prediction methods. The experimental results on the BetaSheet916 data set showed significant improvements in the execution time, the accuracy of ß-strands' alignment and consequently ß-sheet structure prediction accuracy. The results are available at http://conceptsgate.com/BetaSheet.