AllerHybrid: A Hybrid System to Predict the Allergen Using K-mer and Physicochemical Properties

A protein can be an allergen if it is sequentially or structurally similar to other allergens. There are various existing allergen prediction tools. Most existing allergen prediction tools are either alignment-based or alignment-free. However, none of the existing tools is based on both except the AllerCatPro. Nevertheless, the accuracy of AllerCatPro is very low, which is only 84%. That is why we recommended a hybrid model based on Artificial Neural Network (ANN) and SVM using k-mer of protein sequences and physicochemical properties of amino acids, which can detect the allergen protein 94.28% accurately. It can also detect allergenicity of short length and distantly related protein.