Applications of cuckoo search and ant lion optimization for analyzing protein-protein interaction through regularized Markov clustering on coronavirus

Abstract

All living viruses have important structures such as protein. Proteins can interact with each other forming large networks of Protein-Protein Interaction (PPI). In order to facilitate the study of these PPI networks, there needs to be clustering analysis of the PPI. In this research, we use PPI network datasets from SARS-CoV-2 and humans. The interactions of the PPI network will then be formed into graphs. Regularized Markov Clustering (RMCL) is used to perform graph clustering. RMCL consists of three main steps which are regularization, inflation, and pruning. The RMCL algorithm is a variant of Markov Clustering (MCL). However, the inflation parameter in RMCL must be inputted manually by the user to obtain the best results. To solve the limitations of RMCL, we developed a new method by combining each Cuckoo Search (CS) and Ant Lion Optimization (ALO) with the original RMCL algorithm. The optimizers are used to optimize the inflation parameter in RMCL. CS and ALO are a part of swarm intelligence which is inspired by the behaviour of cuckoo birds and antlions in nature. The results show that the interactions formed from CS-RMCL vary from 1401 to 1402. It is more stable than the interactions formed from ALO-RMCL which ranges from 1408 to 3641. The difference between the best elite in each iteration of ALO-RMCL is very influential to the interaction compared to the best nest from the CS-RMCL.