Linear B Cell Epitope Prediction using ProtVec based Encoder and Feed Forward Neural Network

Beetleis an extensive framework for linear b cell epitope prediction. It is a novel deep learning-oriented multi-task framework aimed at linear B-cell epitope prediction that entails the creation of a neural network model that relies on recurrent layers and Transformer blocks for sequence-based analysis. It is highly extensible, thus, can be modified to accommodate new improvements. This study introduces a new innovative amino acid encoding approach, grounded in ProtVec, to facilitate the model in acquiring a nuanced understanding of epitope representations. The aim is to use the ProtVec embeddings for encoding the sequences for enhancing the prediction using the Beetle framework. The experimental analysis conducted on data sourced from IEDB demonstrate the effectiveness of the proposed upgrade in the framework.