DeepANIS

DeepANIS predicts antibody paratopes from concatenated Complementarity Determining Region (CDR) sequences using bidirectional long-short-term memory (BiLSTM) networks and transformer encoders to model residue dependencies and improve accuracy and interpretability for antibody mechanistic research and design.

Key Features:

• Concatenation of CDR Sequences: Concatenates the sequences of Complementarity Determining Regions (CDRs) within a single antibody to capture nonlocal interactions between different CDRs critical for paratope prediction.
• Integration of BiLSTM and Transformer Networks: Leverages bidirectional long-short-term memory (BiLSTM) networks alongside transformer encoders to model dependencies among residues within the concatenated CDR sequences and enhance interpretability.
• Improved Prediction Accuracy: Has been shown to outperform existing methods in predicting antibody paratopes.

Scientific Applications:

• Antibody Mechanistic Research: Identifies paratope regions to support analysis of molecular interactions between antibodies and antigens.
• Antibody Design: Supports rational design of antibodies with desired binding properties by providing accurate paratope predictions.

Methodology:

Concatenation of antibody CDR sequences followed by modeling with bidirectional long-short-term memory (BiLSTM) networks and transformer encoders to capture dependencies among residues.