X-Meeting presentations

EXPANDING THE STRUCTURAL REPERTOIRE OF ANTIBODIES: IN SILICO APPROACHES FOR DATA ENRICHMENT FOR APPLICATION IN MACHINE LEARNING.

Published in 21/11/2024 - ISBN: 978-65-272-0843-3

Link All Papers
LinkSubmission
Paper Title
EXPANDING THE STRUCTURAL REPERTOIRE OF ANTIBODIES: IN SILICO APPROACHES FOR DATA ENRICHMENT FOR APPLICATION IN MACHINE LEARNING.
Authors
  • Diego da Silva de Almeida
  • Geraldo Rodrigues Sartori
  • Matheus do Vale Almeida
  • EDUARDO MENEZES GAIETA
  • Herqüimedes Glaudys da Silva Avelino
  • Jean Vieira Sampaio
  • Francisco Flávio de Assunção Rabelo
  • Andrielly Henriques dos Santos Costa
  • João Herminio Martins Da Silva
Modality
Poster
Subject area
Database and Software Development
Publishing Date
21/11/2024
Country of Publishing
Brazil | Brasil
Language of Publishing
Inglês
Paper Page
https://www.even3.com.br/anais/xmeeting-2024/832076-expanding-the-structural-repertoire-of-antibodies--in-silico-approaches-for-data-enrichment-for-application-in-ma
ISBN
978-65-272-0843-3
Keywords
molecular docking, antibody design, molecular modeling, artificial intelligence, dataset
Summary
Machine learning-based algorithms have been central to therapeutic antibody design, from methods for predicting three-dimensional structures to optimizing antibody-antigen interactions. These algorithms are trained on sequence and/or structure datasets. While antibody sequence databases have extensive repertoires, structural data remains limited, especially on antibody-antigen complexes, hindering the construction of models based solely on structures. In this study, we aimed to generate antibody-antigen complexes in silico to exponentially expand the number of structures available in current databases. To achieve this, starting from AbSet, an internal dataset containing standardized antibody-antigen complexes determined experimentally and retrieved from the Protein Data Bank (PDB), two new subsets were generated. The first one is comprised by submitting each complex to molecular redocking with the HADDOCK software, generating 250 new binding poses for each complex. The second subset was established by modeling the antibodies and their target antigens identified from the AbSet. The variable domains of the antibodies are modeled using the ABodyBuilder2 software, while the structures of the modeled antigens are obtained from the AlphaFold database. The antibody-antigen complexes were constructed through molecular docking. The poses obtained by the two subsets were classified into four categories based on their DockQ values, which can be high quality, medium quality, acceptable, and incorrect. Additionally, molecular descriptors were applied to extract the biochemical characteristics of the antibody-antigen complexes. As a result, an extensive and diverse dataset comprising over 300,000 poses of antibody-antigen complexes was generated. This dataset incorporated molecular descriptors representing the characteristics of amino acid residues and their surroundings, capable of capturing subtle modifications in antibody-antigen interactions. This approach resulted in a robust repository of reliable decoys for antibody-antigen interactions, providing invaluable resources to enhance structure-based AI algorithms. Thus, this work provided valuable insights and datasets for future research on antibody-antigen interactions.
Title of the Event
20º Congresso Brasileiro de Bioinformática: X-Meeting 2024
City of the Event
Salvador
Title of the Proceedings of the event
X-Meeting presentations
Name of the Publisher
Even3
Means of Dissemination
Meio Digital

How to cite

ALMEIDA, Diego da Silva de et al.. EXPANDING THE STRUCTURAL REPERTOIRE OF ANTIBODIES: IN SILICO APPROACHES FOR DATA ENRICHMENT FOR APPLICATION IN MACHINE LEARNING... In: X-Meeting presentations. Anais...Salvador(BA) Hotel Deville Prime, 2024. Available in: https//www.even3.com.br/anais/xmeeting-2024/832076-EXPANDING-THE-STRUCTURAL-REPERTOIRE-OF-ANTIBODIES--IN-SILICO-APPROACHES-FOR-DATA-ENRICHMENT-FOR-APPLICATION-IN-MA. Access in: 27/05/2025

Paper

Even3 Publicacoes