An estimated 170 million people, approximately 3% of the world population, are chronically infected with the hepatitis C virus (HCV). More than 350 000 deaths are reported annually, which are caused by HCV. HCV, similar to a variety of viruses, causes disease in humans by altering protein–protein interactions within the host cells. Experimental approaches for the detection of host–virus PPIs have many inherent limitations. Computational approaches to predict these interactions are therefore of significant importance. While many studies have been developed to predict intra-species PPIs in the last decade, predictions on inter-species PPIs such as human–HCV PPIs are rare. In this study, we developed an ensemble learning method to predict PPIs between human and HCV proteins. Our model utilises four well-established diverse learners as base classifiers including random forest (RF), Naïve Bayes (NB), support vector machine (SVM) and multilayer perceptron (MLP). In addition, an MLP was used as a meta-learner to combine base learners' predictions to provide the final prediction. To encode human and HCV proteins as feature vectors, we used six different descriptors as follows: amino acid composition (ACC), pseudo amino acid composition (PAC), evolutionary information feature, network centrality measures, tissue information and post-translational modification information. To assess the prediction power of the proposed method, we assembled a benchmark dataset composed of confident positive and negative PPIs. In a 10-fold cross-validation experiment, our prediction method achieved accuracy and specificity as high as 83% and 94%, respectively. Furthermore, in an independent test set the proposed method achieved an accuracy of 84% and a specificity of 92%. When compared with the existing method, our method showed a better performance. These results revealed that our method is suitable for performing PPI prediction in a host–pathogen context.

Early diagnosis of lung cancers and distinction between the tumor types (Small Cell Lung Cancer (SCLC) and Non-Small Cell Lung Cancer (NSCLC) are very important to increase the survival rate of patients. Herein, we propose a diagnostic system based on sequence-derived structural and physicochemical attributes of proteins that involved in both types of tumors via feature extraction, feature selection and prediction models. 1497 proteins attributes computed and important features selected by 12 attribute weighting models and finally machine learning models consist of seven SVM models, three ANN models and two NB models applied on original database and newly created ones from attribute weighting models; models accuracies calculated through 10-fold cross and wrapper validation (just for SVM algorithms). In line with our previous findings, dipeptide composition, autocorrelation and distribution descriptor were the most important protein features selected by bioinformatics tools. The algorithms performances in lung cancer tumor type prediction increased when they applied on datasets created by attribute weighting models rather than original dataset. Wrapper-Validation performed better than X-Validation; the best cancer type prediction resulted from SVM and SVM Linear models (82%). The best accuracy of ANN gained when Neural Net model applied on SVM dataset (88%). This is the first report suggesting that the combination of protein features and attribute weighting models with machine learning algorithms can be effectively used to predict the type of lung cancer tumors (SCLC and NSCLC).

Khoram NM, et al. J Breast Cancer. 2016 Mar;19(1):18-25. https://doi.org/10.4048/jbc.2016.19.1.18

Human serum albumin (HSA) is a soluble blood protein which binds to small molecules (such as drugs and toxins) and transfers them within the blood circulation. In this research, the interaction of diazinon, as a toxic organophosphate, with HSA was investigated. Various biophysical methods such as fluorescence, ultraviolet–visible (UV-vis), Fourier transform infrared spectroscopy, and molecular docking were utilized to characterize the binding properties of diazinon to HSA under physiological-like condition. The UV-vis spectroscopy showed that the absorption increased and the fluorescence intensity of HSA decreased regularly with regard to the gradual increases of the concentrations of diazinon. Due to the binding constant of ( k_a = 3.367 × 10⁺⁴ M⁻¹), the α-helix structure for the first day and 35 days of incubation were obtained 66.09–55.4% and 59.99–46.48%, respectively, and their amounts in other secondary structures (β-sheet, β-anti, and random (r) coils) were increased. The molecular docking revealed a good binding site in HSA (Trp-214) for diazinon which was related to the considerable alterations in HSA secondary and tertiary structures. There is a close relationship between the secondary structure of protein and its biological activity and after 35 days of incubation, the high toxic concentrations of diazinon can make HSA to be partially unfolded and lose its structure.