Estimating Change in Foldability Due to Multipoint Deletions in Protein Structures.

Insertions/deletions of amino acids in the protein backbone potentially result in altered structural/functional specifications. They can either contribute positively to the evolutionary process or can result in disease conditions. Despite being the second most prevalent form of protein modification, there are no databases or computational frameworks that delineate harmful multipoint deletions (MPD) from beneficial ones. We introduce a positive unlabeled learning-based prediction framework (PROFOUND) that utilizes fold-level attributes, environment-specific properties, and deletion site-specific properties to predict the change in foldability arising from such MPDs, both in the non-loop and loop regions of protein structures. In the absence of any protein structure dataset to study MPDs, we introduce a dataset with 153 MPD instances that lead to native-like folded structures and 7650 unlabeled MPD instances whose effect on the foldability of the corresponding proteins is unknown. PROFOUND on 10-fold cross-validation on our newly introduced dataset reports a recall of 82.2% (86.6%) and a fall out rate (FR) of 14.2% (20.6%), corresponding to MPDs in the protein loop (non-loop) region. The low FR suggests that the foldability in proteins subject to MPDs is not random and necessitates unique specifications of the deleted region. In addition, we find that additional evolutionary attributes contribute to higher recall and lower FR. The first of a kind foldability prediction system owing to MPD instances and the newly introduced dataset will potentially aid in novel protein engineering endeavors.

A two-stage approach towards protein secondary structure classification.

Protein secondary structure (PSS) describes the local folded structures which get formed inside a polypeptide due to interactions among atoms of the backbone. Generally, globular proteins are divided into four classes, namely all-α, all-ß, α + ß, and α/ß. As nearly 90% of proteins fall into the said four classes, these are mostly considered for the purpose of computational classification of proteins. Classification of PSS is important for different biological functions that include protein fold recognition, tertiary structure prediction, prediction of DNA-binding sites, and reduction of the conformation search space among others. In this paper, we have proposed a machine learning-based model for secondary structure classification of proteins into four classes: all-α, all-ß, α + ß, and α/ß. In doing so, we have considered both sequence-based and structure-based features. At first, mutual information (MI), a filter-based feature selection method, is used to remove the redundant features, and then these selected features are used to train three different classifiers-random forest, K-nearest neighbor (KNN), and multi-layer perceptron (MLP). After that, some standard classifier combination approaches are applied to integrate the decision made by the said classifiers and it has been found that weighted product rule performs the best among all. The overall accuracies obtained using the proposed model on the four standard datasets, namely 640, 1189, 25pdb, and fc699 are 86.89%, 92.93%, 91.38%, and 94.87% respectively. The proposed model outperforms some state-of-the-art methods considered here for comparison. Significantly high classification accuracy produced by our proposed model on four datasets is attributed to the development of a comprehensive feature set (by eliminating redundant features through feature selection technique) which is then passed through an ensemble consists of three different classifiers. Assigning different weights to the outcome of different classifiers thus proved to be useful in designing the model for predicting the secondary structure of proteins based on its sequence-based and structure-based features. Graphical abstract.

Genetic algorithm based cancerous gene identification from microarray data using ensemble of filter methods.

Microarray datasets play a crucial role in cancer detection. But the high dimension of these datasets makes the classification challenging due to the presence of many irrelevant and redundant features. Hence, feature selection becomes irreplaceable in this field because of its ability to remove the unrequired features from the system. As the task of selecting the optimal number of features is an NP-hard problem, hence, some meta-heuristic search technique helps to cope up with this problem. In this paper, we propose a 2-stage model for feature selection in microarray datasets. The ranking of the genes for the different filter methods are quite diverse and effectiveness of rankings is datasets dependent. First, we develop an ensemble of filter methods by considering the union and intersection of the top-n features of ReliefF, chi-square, and symmetrical uncertainty. This ensemble allows us to combine all the information of the three rankings together in a subset. In the next stage, we use genetic algorithm (GA) on the union and intersection to get the fine-tuned results, and union performs better than the latter. Our model has been shown to be classifier independent through the use of three classifiers-multi-layer perceptron (MLP), support vector machine (SVM), and K-nearest neighbor (K-NN). We have tested our model on five cancer datasets-colon, lung, leukemia, SRBCT, and prostate. Experimental results illustrate the superiority of our model in comparison to state-of-the-art methods. Graphical abstract ᅟ.