A novel machine learning method for cytokine-receptor interaction prediction

Abstract

Most essential functions are associated with various protein-protein interactions, particularly the cytokine-receptor interaction. Knowledge of the heterogeneous network of cytokine- receptor interactions provides insights into various human physiological functions. However, only a few studies are focused on the computational prediction of these interactions. In this study, we propose a novel machine-learning-based method for predicting cytokine-receptor interactions. A protein sequence is first transformed by incorporating the sequence evolutional information and then formulated with the following three aspects: (1) the k-skip-n-gram model, (2) physicochemical properties, and (3) local pseudo position-specific score matrix (local PsePSSM). The random forest classifier is subsequently employed to predict potential cytokine-receptor interactions. Experimental results on a dataset of Homo sapiens show that the proposed method exhibits improved performance, with 3.4% higher overall prediction accuracy, than existing methods.

Exploring local discriminative information from evolutionary profiles for cytokine–receptor interaction prediction

“…We proposed a machine learning based prediction method featured by the 828-Dhybrid features using the Random Forest (RF) classifier. As reported in [28], it achieved an overall accuracy of 83.7% with 10-fold cross validation on the dataset that contains 203 actual cytokine-receptor interactions and 203 non-cytokine-receptor interactions.…”

“…In particular, a bunch of recent work has reported its high efficiency when applied into various fields, such as protein structural class prediction [52], DNA-binding protein prediction [59], as well as cytokine-receptor interaction prediction [28], and cell penetrating peptide prediction [60], etc.…”

“…The construction procedure of this new dataset is similar with that of the dataset in our previous study [28]. For positive dataset construction, positive data (known cytokine-receptor interactions) were downloaded from the Kyoto Encyclopedia of Genes and Genomes (pathway: map04060) [63].…”

“…Hybrid methods usually fuse different kinds of feature types (e.g., sequence-based, profile-based). For instance, in our previous study [28], we proposed a 828-D (dimensional) hybrid feature vector from multi-views: the k-skip-n-gram model, physicochemical model and PSSM-based model. For the classification algorithm, association rule based classifier [29], Support Vector Machine (SVM) [30][31][32][33][34][35][36][37][38] and Random Forest (RF) [39][40][41][42] are widely used in the bioinformatics discriminative problem.…”

“…It is a SVM-based method that uses the 188-D physicochemical feature vector to make predictions. The other one is the method we proposed in our previous study [28]. We proposed a machine learning based prediction method featured by the 828-Dhybrid features using the Random Forest (RF) classifier.…”