An Efficient Dilated DTCN with Residual LSTM-based Allergenicity Prediction for Chemical Compounds using Optimal Weighted Feature Selection and Improved Optimization Mechanism

Allergy is the sudden response of the immunity system which arises later the disclosure of allergens like chemicals, proteins and peptides. In the early times, multiple techniques were formed for allergenicity forecasting of peptides and proteins. However, there is no technique to anticipate the chemical and potential allergenic. The standard food allergen recognition mainly depends on vitro and vivo research, which frequently needs more time and is expensive. Artificial Intelligence (AI) operates a quick food allergen detection model, which solves the above-defined problems and becomes an effective auxiliary tool. Evaluation of potential allergenicity of protein is required at any movement of transgenic proteins implemented in the food chain. Therefore, it is necessary to solve more complications involved in the conventional allergenicity prediction technique for chemical compounds. Thus, a new allergenicity prediction framework for chemical compounds is designed using deep learning techniques. Initially, essential allergenicity data are accumulated from the standard measures. Next, the acquired data are presented for the optimal weighted feature selection stage. In this phase, the weights and features are selected optimally using Opposition Tasmanian Devil Optimization (OTDO). Further, the adopted optimal weighted features are offered to the allergenicity forecasting stage. Here, the allergenicity is predicated on using the developed "Dilated Deep Temporal Context Networks with Residual Long Short-Term Memory Networks (DiDNet-ResLSTM)". Hence, the recommended allergenicity prediction model effectively higher performance rate in different experimental observations.