Interrelating N-gram based Protein Sequences using LSTMs with Parallel Capsule Routing

Genome data is high-dimensional, complex data that demands effective data management policies for high-performance results to develop efficient solutions. The protein perspective of this data deals with the proteome, and their basic information to be conceptually categorized, such as the Protein-Protein Interactions, structure, functions, and its processing events. Analyzing the interaction of proteins is vital to studying cell cycle and gene expression regulation, including energy and substance metabolism processes in living bodies, and help understand the role of proteins and their functional connections in biological systems. In this paper, precise ontological development to establish a solid framework for the analysis of protein interactions is performed using Long Short-Term Memory Model with parallel capsule routing. The polypeptide strings are converted into N-grams and LSTMs help co-relate the N-gram motifs within an amino acid sequence. The interactions between the N-gram sequences of each input protein are modeled using capsules and the overall interaction between any two parallel protein sequences is mapped through dynamic routing. The implemented model has performed at a score of 92.1 % accuracy with the Area Under the Curve value at 96.2%.