Molecular k-mer Images with Vision Transformers for Blood–Brain Barrier Permeability Prediction

• Giovanna A. P. Soares UFRN
• Hannah I. da S. Marques UFRN
• Raquel de M. Barbosa UFRN
• Marcelo A. C. Fernandes UFRN

DOI: https://doi.org/10.5753/eniac.2026.28684

Resumo

A predição da permeabilidade da barreira hematoencefálica (BBB) é um problema importante no desenvolvimento de fármacos, pois determina se uma molécula consegue alcançar o sistema nervoso central. Este trabalho propõe uma abordagem baseada na geração de representações visuais moleculares derivadas de sequências SMILES. Inicialmente, as moléculas são tokenizadas e utilizadas para construir matrizes de contagem de k-mers (MC), posteriormente convertidas em imagens denominadas Molecule Image Representations (MIR). Essas imagens são processadas por um modelo Vision Transformer (ViT), responsável por extrair embeddings moleculares utilizados por classificadores supervisionados. Experimentos conduzidos nos datasets BBBP e B3DB demonstraram que a abordagem proposta produz resultados competitivos em termos de AUC, alcançando 0,9281 no BBBP (SVM) e 0,9125 no B3DB (RF). Os resultados indicam que os embeddings gerados pelo ViT capturam características estruturais relevantes das moléculas. Além disso, o ajuste fino parcial do modelo contribui para melhorar a qualidade das representações aprendidas. De forma geral, o método apresenta uma alternativa simples e eficaz às abordagens baseadas em grafos moleculares e descritores físico-químicos extensivos.

Referências

Breiman, L. (2001). Random forests. Machine learning, 45(1):5–32.

Chen, T. and Guestrin, C. (2016). XGBoost: A scalable tree boosting system. In Proceedings of the 22nd ACM SIGKDD international conference on knowledge discovery and data mining, pages 785–794.

Cortes, C. and Vapnik, V. (1995). Support-vector networks. Machine learning, 20(3):273–297.

Coutinho, M. G. F., Câmara, G. B. M., Barbosa, R. d. M., and Fernandes, M. A. C. (2023). Sars-cov-2 virus classification based on stacked sparse autoencoder. Computational and Structural Biotechnology Journal, 21:284–298.

Câmara, G. B. M., Coutinho, M. G. F., Silva, L. M. D. d., Gadelha, W. V. d. N., Torquato, M. F., Barbosa, R. d. M., and Fernandes, M. A. C. (2022). Convolutional neural network applied to sars-cov-2 sequence classification. Sensors, 22(15):5730.

De Souza, J. G., Fernandes, M. A., and de Melo Barbosa, R. (2022). A novel deep neural network technique for drug–target interaction. Pharmaceutics, 14(3):625.

Dosovitskiy, A., Beyer, L., Kolesnikov, A., Weissenborn, D., Zhai, X., Unterthiner, T., Dehghani, M., Minderer, M., Heigold, G., Gelly, S., et al. (2021). An image is worth 16x16 words: Transformers for image recognition at scale. International Conference on Learning Representations (ICLR).

Dou, B., Zhu, Z., Merkurjev, E., Ke, L., Chen, L., Jiang, J., Zhu, Y., Liu, J., Zhang, B., and Wei, G.-W. (2023). Machine learning methods for small data challenges in molecular science. Chemical Reviews, 123(13):8736–8780.

Fabian, B., Edlich, T., Gaspar, H., Segler, M., Meyers, J., Fiscato, M., and Ahmed, M. (2020). Molecular representation learning with language models and domain-relevant auxiliary tasks. arXiv preprint arXiv:2011.13230.

Guzman-Pando, A., Ramirez-Alonso, G., Arzate-Quintana, C., and Camarillo-Cisneros, J. (2024). Deep learning algorithms applied to computational chemistry. Molecular Diversity, 28(4):2375–2410.

Huang, E. T., Yang, J.-S., Liao, K. Y., Tseng, W. C., Lee, C., Gill, M., Compas, C., See, S., and Tsai, F.-J. (2024). Predicting blood–brain barrier permeability of molecules with a large language model and machine learning. Scientific reports, 14(1):15844.

Jafarpour, S., Asefzadeh, M., and Aboutaleb, E. (2024). The application of machine learning in predicting the permeability of drugs across the blood brain barrier. Iranian Journal of Pharmaceutical Research: IJPR, 23(1):e149367.

Jia, H. and Sosso, G. C. (2024). Transparent machine learning model to understand drug permeability through the blood–brain barrier. Journal of Chemical Information and Modeling, 64(23):8718–8728.

Kumar, S., Yadav, S., Vora, D. S., Paul, P., and Dhanjal, J. K. (2025). graphb3—an interpretable graph learning approach for predicting blood–brain barrier permeability. Briefings in Bioinformatics, 26(6):bbaf679.

Meng, F., Xi, Y., Huang, J., and Ayers, P. W. (2021). B3DB: A multiscale dataset for understanding and predicting the blood–brain barrier permeability. Scientific Data, 8(1):289.

Qin, M., Sun, Z., Feng, L., Han, C., Xia, J., and Han, L. (2025). Moleculeformer is a gcntransformer architecture for molecular property prediction. Communications Biology, 8(1):1668.

Tong, X., Wang, D., Ding, X., Tan, X., Ren, Q., Chen, G., Rong, Y., Xu, T., Huang, J., Jiang, H., et al. (2022). Blood–brain barrier penetration prediction enhanced by uncertainty estimation. Journal of Cheminformatics, 14(1):44.

Visan, A. I. and Negut, I. (2024). Integrating artificial intelligence for drug discovery in the context of revolutionizing drug delivery. Life, 14(2):233.

Wu, Z., Ramsundar, B., Feinberg, E. N., Gomes, J., Geniesse, C., Pappu, A. S., Leswing, K., and Pande, V. (2018). Moleculenet: a benchmark for molecular machine learning. Chemical science, 9(2):513–530.

Xiong, Z., Wang, D., Liu, X., Zhong, F., Wan, X., Li, X., Li, Z., Luo, X., Chen, K., Jiang, H., et al. (2019). Pushing the boundaries of molecular representation for drug discovery with the graph attention mechanism. Journal of medicinal chemistry, 63(16):8749–8760.

Yang, K., Swanson, K., Jin, W., Coley, C., Eiden, P., Gao, H., Guzman-Perez, A., Hopper, T., Kelley, B., Mathea, M., et al. (2019). Analyzing learned molecular representations for property prediction. Journal of chemical information and modeling, 59(8):3370–3388.

Zeng, X., Xiang, H., Yu, L., Wang, J., Li, K., Nussinov, R., and Cheng, F. (2022). Accurate prediction of molecular properties and drug targets using a self-supervised image representation learning framework. Nature Machine Intelligence, 4(11):1004–1016.