Similarity-Based Gene Duplication Prediction in Protein-Protein Interaction Using Deep Artificial Ecosystem Network

Abstract

In the living organism, almost entire cell functions are performed by protein-protein interactions. As experimental and computing technology advances, yet more Protein-Protein Interaction (PPI) data becomes processed, and PPI networks become denser. The traditional methods utilize the network structure to examine the protein structure. Still, it consumes more time and cost and creates computing complexity when the system has gene duplications and a complementary interface. This research uses gene expression patterns to introduce a deep artificial ecosystem for gene duplication counting and cancer cell prediction. The main objective of this research is to predict the MYC proteins influence level, which is in charge of controlling cell growth and death in gene expression of lung cancer. Small body parts are responsible for these protein interactions, which are crucial for understanding life's activities. To achieve the research objective, a similarity-based clustering approach is employed for gene duplication counting, and Artificial Ecosystem Optimizer based Minimal Gated Recurrent Unit network (AEOMGRU) network-based approach is introduced to predict the cancer gene patterns. The proposed models' efficiency is compared to recently develop bio-inspired optimizer deep neural network techniques such as GAANN, PSOANN, and classic GRU. The efficiency of the proposed classifier shows the highest concerning the performance metrics weight average accuracy ratio of 99.08%, average.