Tumor Type Classification from Somatic Mutations Using a Transformer-Based Model

Authors

  • Saeid Sanaei Faculty of Electrical and Computer Engineering, Mazandaran University of Science and Technology, Behshahr, Iran.
  • Reza Javanmard Alitappeh * Faculty of Electrical and Computer Engineering, Mazandaran University of Science and Technology, Behshahr, Iran.

https://doi.org/10.22105/ahse.v3i3.67

Abstract

Somatic mutation profiles are sparse and variable in length, which makes tumor-type classification difficult with standard fixed-vector inputs. We propose a transformer-based model that represents each The Cancer Genome Atlas (TCGA) sample as a sequence of mutation tokens combining gene, genomic-position, substitution, sequence-context, variant-type, Variant Allele Fraction (VAF), read-depth, hotspot, and consequence features. Categorical fields were encoded by stable hashing, and numerical fields were normalized using only the training split of each fold. After filtering tumor types with at least 180 samples, 8,910 samples from 20 classes were evaluated. In stratified eight-fold cross-validation, the model achieved 64.4% accuracy, 65.4% weighted accuracy, 95.8% one-vs-rest ROC-AUC, 86.7% Top-3 accuracy, and 93.1% Top-5 accuracy. These results support sequence-based mutation modeling for ranked tumor-type prediction. 

Keywords:

Tumor type classification, Somatic mutations, Transformer model, Deep learning, TCGA, Precision oncology

References

  1. [1] Anaya, J., Sidhom, J. W., Mahmood, F., & Baras, A. S. (2024). Multiple-instance learning of somatic mutations for the classification of tumour type and the prediction of microsatellite status. Nature Biomedical Engineering, 8(1), 57–67. https://doi.org/10.1038/s41551-023-01120-3
  2. [2] Sanjaya, P., Maljanen, K., Katainen, R., Waszak, S. M., Ambrose, J. C., Arumugam, P., … ., & Consortium, G. E. R. (2023). Mutation-Attention (MuAt): Deep representation learning of somatic mutations for tumour typing and subtyping. Genome Medicine, 15(1), 47. https://doi.org/10.1186/s13073-023-01204-4
  3. [3] Jiao, W., Atwal, G., Polak, P., Karlic, R., Cuppen, E., Al-Shahrour, F., … ., & Consortium, P. (2020). A deep learning system accurately classifies primary and metastatic cancers using passenger mutation patterns. Nature Communications, 11(1), 728. https://doi.org/10.1038/s41467-019-13825-8
  4. [4] Zelli, V., Manno, A., Compagnoni, C., Ibraheem, R. O., Zazzeroni, F., Alesse, E., … ., & Tessitore, A. (2023). Classification of tumor types using XGBoost machine learning model: A vector space transformation of genomic alterations. Journal of Translational Medicine, 21(1), 836. https://doi.org/10.1186/s12967-023-04720-4
  5. [5] Kobayashi, K., Bolatkan, A., Shiina, S., & Hamamoto, R. (2020). Fully-connected neural networks with reduced parameterization for predicting histological types of lung cancer from somatic mutations. Biomolecules, 10(9), 1–16. https://doi.org/10.3390/biom10091249
  6. [6] Zheng, W., Pu, M., Li, X., Du, Z., Jin, S., Li, X., … ., & Zhang, Y. (2023). Deep learning model accurately classifies metastatic tumors from primary tumors based on mutational signatures. Scientific Reports, 13(1), 8752. https://doi.org/10.1038/s41598-023-35842-w
  7. [7] Nguyen, L., Van Hoeck, A., & Cuppen, E. (2022). Machine learning-based tissue of origin classification for cancer of unknown primary diagnostics using genome-wide mutation features. Nature communications, 13(1), 4013. https://doi.org/10.1038/s41467-022-31666-w
  8. [8] Moon, I., LoPiccolo, J., Baca, S. C., Sholl, L. M., Kehl, K. L., Hassett, M. J., … ., & Gusev, A. (2023). Machine learning for genetics-based classification and treatment response prediction in cancer of unknown primary. Nature Medicine, 29(8), 2057–2067. https://doi.org/10.1038/s41591-023-02482-6
  9. [9] Darmofal, M., Suman, S., Atwal, G., Toomey, M., Chen, J. F., Chang, J. C., … ., & Morris, Q. (2024). Deep-learning model for tumor-type prediction using targeted clinical genomic sequencing data. Cancer Discovery, 14(6), 1064–1081. https://doi.org/10.1158/2159-8290.CD-23-0996
  10. [10] Huang, Y., Pfeiffer, S. M., & Zhang, Q. (2023). Primary tumor type prediction based on US nationwide genomic profiling data in 13,522 patients. Computational and Structural Biotechnology Journal, 21, 3865–3874. https://doi.org/10.1016/j.csbj.2023.07.036
  11. [11] Danyi, A., Jager, M., & de Ridder, J. (2021). Cancer type classification in liquid biopsies based on sparse mutational profiles enabled through data augmentation and integration. Life, 12(1), 1. https://doi.org/10.3390/life12010001
  12. [12] Conway, A. M., Pearce, S. P., Clipson, A., Hill, S. M., Chemi, F., Slane-Tan, D., … ., & Rothwell, D. G. (2024). A cfDNA methylation-based tissue-of-origin classifier for cancers of unknown primary. Nature Communications, 15(1), 3292. https://doi.org/10.1038/s41467-024-47195-7
  13. [13] Michuda, J., Breschi, A., Kapilivsky, J., Manghnani, K., McCarter, C., Hockenberry, A. J., … ., & Taxter, T. (2023). Validation of a transcriptome-based assay for classifying cancers of unknown primary origin. Molecular Diagnosis & Therapy, 27(4), 499–511. https://doi.org/10.1007/s40291-023-00650-5
  14. [14] Ziegler, J., Hechtman, J. F., Rana, S., Ptashkin, R. N., Jayakumaran, G., Middha, S., … ., & Zehir, A. (2025). A deep multiple instance learning framework improves microsatellite instability detection from tumor next generation sequencing. Nature Communications, 16(1), 136. https://doi.org/10.1038/s41467-024-54970-z
  15. [15] Chen, J., Wang, M., Zhao, D., Li, F., Wu, H., Liu, Q., & Li, S. (2023). MSINGB: A novel computational method based on NGBoost for identifying microsatellite instability status from tumor mutation annotation data. Interdisciplinary Sciences: Computational Life Sciences, 15(1), 100–110. https://doi.org/10.1007/s12539-022-00544-w

Downloads

Published

2026-07-10

Issue

Vol. 3 No. 3 (2026): Annals of Healthcare Systems Engineering

Section

Articles

How to Cite

Sanaei, S., & Javanmard Alitappeh, R. (2026). Tumor Type Classification from Somatic Mutations Using a Transformer-Based Model. Annals of Healthcare Systems Engineering, 3(3), 159-170. https://doi.org/10.22105/ahse.v3i3.67

Similar Articles

1-10 of 38

You may also start an advanced similarity search for this article.