Integração de Parâmetros Físico-Químicos e Machine Learning para a Identificação de Epítopos
DOI:
https://doi.org/10.5540/03.2026.012.01.0296Keywords:
Sistema Imune, Imunoterapia, Antígenos, Random Forest, ClassificaçãoAbstract
Este estudo propõe um método para classificar epítopos em três categorias: MHC classe I, MHC classe II e epítopos de células B, utilizando dados do Immune Epitope Database (IEDB) e técnicas de aprendizado de máquina. A classificação de epítopos é importante para o desenvolvimento de vacinas e terapias imunológicas, pois permite identificar alvos específicos e prever respostas imunes. Parâmetros físico-químicos foram usados para representar as sequências de aminoácidos. O modelo Random Forest (RF) obteve os melhores resultados, com acurácia de 84,32% e F1-score macro de 0,8276, destacando-se na classificação de MHC I (F1-score de 0,9557).
Downloads
References
J. Bergstra e Y. Bengio. “Random Search for Hyper-Parameter Optimization”. Em: Journal of Machine Learning Research 13.10 (2012), pp. 281–305.
C. M. Bishop. Pattern Recognition and Machine Learning (Information Science and Statistics). Berlin, Heidelberg: Springer-Verlag, 2006. isbn: 0387310738.
D. S. Chen e I. Mellman. “Elements of cancer immunity and the cancer–immune set point”. Em: Nature 541.7637 (2017), pp. 321–330. doi: 10.1038/nature21349.
K. K. Jensen et al. “Improved methods for predicting peptide binding affinity to MHC class II molecules”. Em: Immunology 154.3 (2018), pp. 394–406. doi: 10.1111/imm.12889.
M. C. Jespersen et al. “BepiPred-2.0: improving sequence-based B-cell epitope prediction using conformational epitopes”. Em: Nucleic Acids Research 45.W1 (2017), pp. W24–W29. doi: 10.1093/nar/gkx346.
C. H. June et al. “CAR T cell immunotherapy for human cancer”. Em: Science 359.6382 (2018), pp. 1361–1365. doi: 10.1126/science.aar6711.
R. Kohavi. “A study of cross-validation and bootstrap for accuracy estimation and model selection”. Em: Proceedings of the 14th International Joint Conference on Artificial Intelligence - Volume 2. San Francisco, CA, USA: Morgan Kaufmann Publishers Inc., 1995, pp. 1137–1143.
E. E. G. Kozlova et al. “Classification epitopes in groups based on their protein family”. Em: BMC Bioinformatics 16.19 (2015), pp. S7. doi: 10.1186/1471-2105-16-S19-S7.
P. A. Ott et al. “A Phase Ib Trial of Personalized Neoantigen Therapy Plus Anti-PD-1 in Patients with Advanced Melanoma, Non-small Cell Lung Cancer, or Bladder Cancer”. Em: Cell 183.2 (2020), pp. 347–362. doi: 10.1016/j.cell.2020.08.053.
M. Rosenzwajg et al. “Immunological and clinical effects of low-dose interleukin-2 across 11 autoimmune diseases in a single, open clinical trial”. Em: Annals of the Rheumatic Diseases 78.2 (2019), pp. 209–217. doi: 10.1136/annrheumdis-2018-214229.
U. Sahin et al. “COVID-19 vaccine BNT162b1 elicits human antibody and TH1 T cell responses”. Em: Nature 586.7830 (2020), pp. 594–599. doi: 10.1038/s41586-020-2814-7.
J. L. Sanchez-Trincado, M. Gomez-Perosanz e P. A. Reche. “Fundamentals and Methods for T- and B-Cell Epitope Prediction”. Em: Journal of Immunology Research 2017.1 (2017), pp. 2680160. doi: 10.1155/2017/2680160.
A. Sette e R. Rappuoli. “Reverse Vaccinology: Developing Vaccines in the Era of Genomics”. Em: Immunity 33.4 (2010), pp. 530–541. doi: 10.1016/j.immuni.2010.09.017.
Z. Tayebi, S. Ali e M. Patterson. “TCellR2Vec: efficient feature selection for TCR sequences for cancer classification”. Em: PeerJ Computer Science 10 (2024), pp. e2239. doi: 10.7717/peerj-cs.2239.
R. Vita et al. “The Immune Epitope Database (IEDB): 2018 update”. Em: Nucleic Acids Research 47.D1 (2018), pp. D339–D343. doi: 10.1093/nar/gky1006.
D. C. Wraith. “Therapeutic peptide vaccines for treatment of autoimmune diseases”. Em: Immunology Letters 122.2 (2009). Special Section: Vaccination Immunology: Prevention and Beyond, pp. 134–136. doi: 10.1016/j.imlet.2008.11.013.
