Powered by OpenAIRE graph
Found an issue? Give us feedback
image/svg+xml Jakob Voss, based on art designer at PLoS, modified by Wikipedia users Nina and Beao Closed Access logo, derived from PLoS Open Access logo. This version with transparent background. http://commons.wikimedia.org/wiki/File:Closed_Access_logo_transparent.svg Jakob Voss, based on art designer at PLoS, modified by Wikipedia users Nina and Beao Estudo Geralarrow_drop_down
image/svg+xml Jakob Voss, based on art designer at PLoS, modified by Wikipedia users Nina and Beao Closed Access logo, derived from PLoS Open Access logo. This version with transparent background. http://commons.wikimedia.org/wiki/File:Closed_Access_logo_transparent.svg Jakob Voss, based on art designer at PLoS, modified by Wikipedia users Nina and Beao
Estudo Geral
Master thesis . 2025
Data sources: Estudo Geral
addClaim

Harnessing TCR repertoire to improve immunogenicity prediction

Authors: Correia, Paulo Alexandre da Costa;

Harnessing TCR repertoire to improve immunogenicity prediction

Abstract

A imunogenicidade nas proteínas terapêuticas representa um desafio significativo no desenvolvimento de novos fármacos, levando à redução da eficácia e a reações adversas nos pacientes. A previsão rigorosa da imunogenicidade in silico é crucial para mitigar riscos durante as fases iniciais da otimização/seleção do composto líder. Esta dissertacao apresenta uma nova framework concebida para melhorar a previsão da imunogenicidade induzida pelas células T, modelando as suas duas etapas principais: a apresentação de péptidos pelas moléculas do Complexo Principal de Histocompatibilidade (MHC) e o subsequente reconhecimento do complexo peptido-MHC (pMHC) por um recetor de células T (TCR).Para tal, foi desenvolvido um modelo unificado de deep learning para a previsão da apresentação pan-alélica em MHCs da classe I e II, demonstrando uma melhoria estatisticamente significativa no desempenho relativamente às ferramentas de ponta existentes (+11.17\% AP). Foram desenvolvidos dois modelos para estudar a ativação do TCR: um “Modelo Clássico” para prever interações específicas pMHC-TCR e um novo “Modelo Agnóstico” que prevê o potencial imunogénico geral sem exigir a entrada do TCR. Estes modelos utilizam arquiteturas avançadas, incluindo embeddings de protein Language Models (pLM) e mecanismos de atenção, para capturar interações biológicas complexas.A framework integrada demonstrou uma correlação positiva com as taxas de anticorpos antidrogas (ADA) observadas clinicamente e identificou com sucesso o único epítopo imunodominante num caso de teste do perfil imunogénico de Natalizumab. Este trabalho fornece uma metodologia robusta e validada que melhora a previsão da imunogenicidade e oferece uma ferramenta promissora para o design racional e a desimunização de terapêuticas proteicas de última geração.

The immunogenicity of therapeutic proteins represents a significant challenge in drug development, potentially leading to reduced efficacy and adverse patient reactions. Accurate in silico prediction of immunogenicity is crucial for mitigating these risks during the early stages of lead optimization. This thesis presents a novel computational framework designed to improve the prediction of T-cell-driven immunogenicity by modeling its two core mechanistic steps: the presentation of peptides by Major Histocompatibility Complex (MHC) molecules and the subsequent recognition of the peptide-MHC (pMHC) complex by a T-cell receptor (TCR).A unified deep learning model for pan-allelic MHC class I and II presentation prediction was developed, which demonstrated a statistically significant improvement in performance over existing state-of-the-art tools (+11.17\% AP). Two models to study the TCR activation were developed: a "Classical Model" for predicting specific pMHC-TCR interactions and a novel "Agnostic Model" that predicts a the general immunogenic potential without requiring the TCR input. These models leverage advanced architectures, including protein language model embeddings and attention mechanisms, to capture complex biological interactions.The integrated framework showed a positive correlation with clinically observed Anti-Drug Antibody (ADA) rates and successfully identified the sole immunodominant epitope in a case study of Natalizumab. This work provides a robust, validated methodology that enhances immunogenicity prediction and offers a promising tool for the rational design and deimmunization of next-generation protein therapeutics.

Dissertação de Mestrado em Engenharia e Ciência de Dados apresentada à Faculdade de Ciências e Tecnologia

Country
Portugal
Related Organizations
Keywords

Machine Learning, Deep Learning, T-cell Receptor, Therapeutic Proteins, Recetor das Células T, Proteínas Terapeuticas, Imunogenicity, Imunogenicidade

  • BIP!
    Impact byBIP!
    selected citations
    These citations are derived from selected sources.
    This is an alternative to the "Influence" indicator, which also reflects the overall/total impact of an article in the research community at large, based on the underlying citation network (diachronically).
    0
    popularity
    This indicator reflects the "current" impact/attention (the "hype") of an article in the research community at large, based on the underlying citation network.
    Average
    influence
    This indicator reflects the overall/total impact of an article in the research community at large, based on the underlying citation network (diachronically).
    Average
    impulse
    This indicator reflects the initial momentum of an article directly after its publication, based on the underlying citation network.
    Average
Powered by OpenAIRE graph
Found an issue? Give us feedback
selected citations
These citations are derived from selected sources.
This is an alternative to the "Influence" indicator, which also reflects the overall/total impact of an article in the research community at large, based on the underlying citation network (diachronically).
BIP!Citations provided by BIP!
popularity
This indicator reflects the "current" impact/attention (the "hype") of an article in the research community at large, based on the underlying citation network.
BIP!Popularity provided by BIP!
influence
This indicator reflects the overall/total impact of an article in the research community at large, based on the underlying citation network (diachronically).
BIP!Influence provided by BIP!
impulse
This indicator reflects the initial momentum of an article directly after its publication, based on the underlying citation network.
BIP!Impulse provided by BIP!
0
Average
Average
Average
Upload OA version
Are you the author of this publication? Upload your Open Access version to Zenodo!
It’s fast and easy, just two clicks!