Interpretable Machine Learning for Identifying Minimal Gene Expression Signatures Associated with Leukemia Subtypes
Keywords:
Leukemia, Gene Expression, Machine Learning, Feature Selection, Molecular Classification, BioinformaticsAbstract
Leukemias can be subdivided into distinct molecular programs, although high dimensional data in the gene-expression domain makes the identification of compact and interpretable diagnostic signatures difficult. In this study, the authors present an interpretable machine-learning workflow to discriminate B-lineage acute lymphoblastic leukemia (B-ALL), T-cell ALL (T-ALL) and AML using the public expression-profiling dataset GSE7186. After quality control, the analysis was performed on 121 diagnostic samples and 5,000 expression features. Leakage-safe SelectKBest filtering was integrated in repeated stratified five-fold cross validation, along with L1 logistic regression, linear support vector machines and Extra Trees, across panels of 5, 10, 20, 50 and 100 features. The highest macro-F1 score (0.681) was obtained by L1 logistic regression using 100 features, which had a balanced accuracy of 0.690 and accuracy of 0.808. A set of 65 recurrent expression probes were identified by a stability analysis, of which 11 were present in all resampling folds. AML and T-ALL were more similar and B-ALL was more easily classified. Results reveal that compact stable transcriptomic signatures can maintain meaningful subtype-discrimination capacity while enhancing the transparency of the model and the interpretation in a biology-oriented leukemia stratification.
References
1.Alaggio, R., Amador, C., Anagnostopoulos, I., et al. (2022). The 5th edition of the World Health Organization Classification of Haematolymphoid Tumours: Lymphoid Neoplasms. Leukemia, 36(7), 1720–1748. https://doi.org/10.1038/s41375-022-01620-2
2.Andersson, A., Ritz, C., Lindgren, D., Edén, P., Lassen, C., Heldrup, J., Olofsson, T., Råde, J., Fontes, M., Porwit-MacDonald, A., Behrendtz, M., Höglund, M., Johansson, B., & Fioretos, T. (2007). Microarray-based classification of a consecutive series of 121 childhood acute leukemias: Prediction of leukemic and genetic subtype as well as of minimal residual disease status. Leukemia, 21, 1198–1203. https://doi.org/10.1038/sj.leu.2404688
3.Arber, D. A., Orazi, A., Hasserjian, R. P., et al. (2022). International Consensus Classification of Myeloid Neoplasms and Acute Leukemias: Integrating morphologic, clinical, and genomic data. Blood, 140(11), 1200–1228. https://doi.org/10.1182/blood.2022015850
4.Barredo Arrieta, A., Díaz-Rodríguez, N., Del Ser, J., et al. (2020). Explainable artificial intelligence (XAI): Concepts, taxonomies, opportunities and challenges toward responsible AI. Information Fusion, 58, 82–115. https://doi.org/10.1016/j.inffus.2019.12.012
5.Beder, T., Hansen, B.-T., Hartmann, A. M., et al. (2023). The gene expression classifier ALLCatchR identifies B-cell precursor ALL subtypes and underlying developmental trajectories across age. HemaSphere, 7(9), e939. https://doi.org/10.1097/HS9.0000000000000939
6.Chicco, D., & Jurman, G. (2020). The advantages of the Matthews correlation coefficient (MCC) over F1 score and accuracy in binary classification evaluation. BMC Genomics, 21, 6. https://doi.org/10.1186/s12864-019-6413-7
7.Greener, J. G., Kandathil, S. M., Moffat, L., & Jones, D. T. (2022). A guide to machine learning for biologists. Nature Reviews Molecular Cell Biology, 23, 40–55. https://doi.org/10.1038/s41580-021-00407-0
8.Hehr, M., Sadafi, A., Matek, C., Lienemann, P., Pohlkamp, C., Haferlach, T., Spiekermann, K., & Marr, C. (2023). Explainable AI identifies diagnostic cells of genetic AML subtypes. PLOS Digital Health, 2(3), e0000187. https://doi.org/10.1371/journal.pdig.0000187
9.Khoury, J. D., Solary, E., Abla, O., et al. (2022). The 5th edition of the World Health Organization Classification of Haematolymphoid Tumours: Myeloid and Histiocytic/Dendritic Neoplasms. Leukemia, 36(7), 1703–1719. https://doi.org/10.1038/s41375-022-01613-1
10.Liu, Z., & Elcheva, I. (2022). A six-gene prognostic signature for both adult and pediatric acute myeloid leukemia identified with machine learning. American Journal of Translational Research, 14(9), 6210–6221.
11.Lundberg, S. M., & Lee, S.-I. (2017). A unified approach to interpreting model predictions. Advances in Neural Information Processing Systems, 30, 4765–4774. https://proceedings.neurips.cc/paper/2017/hash/8a20a8621978632d76c43dfd28b67767-Abstract.html
12.Malta, T. M., Sokolov, A., Gentles, A. J., et al. (2018). Machine learning identifies stemness features associated with oncogenic dedifferentiation. Cell, 173(2), 338–354.e15. https://doi.org/10.1016/j.cell.2018.03.034
13.Morabito, F., Adornetto, C., Monti, P., et al. (2023). Genes selection using deep learning and explainable artificial intelligence for chronic lymphocytic leukemia predicting the need and time to therapy. Frontiers in Oncology, 13, 1198992. https://doi.org/10.3389/fonc.2023.1198992
14.Mosquera Orgueira, A., Peleteiro Raíndo, A., Cid López, M., et al. (2021). Personalized survival prediction of patients with acute myeloblastic leukemia using gene expression profiling. Frontiers in Oncology, 11, 657191. https://doi.org/10.3389/fonc.2021.657191
15.Mosquera Orgueira, A., Peleteiro Raíndo, A., Díaz Arias, J. Á., et al. (2022). Evaluation of the Stellae-123 prognostic gene expression signature in acute myeloid leukemia. Frontiers in Oncology, 12, 968340. https://doi.org/10.3389/fonc.2022.968340
16.National Center for Biotechnology Information. (2007). GSE7186: Microarray-based classification of a consecutive series of 121 childhood acute leukemias [Data set]. Gene Expression Omnibus. https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE7186
17.Nogueira, S., Sechidis, K., & Brown, G. (2018). On the stability of feature selection algorithms. Journal of Machine Learning Research, 18(174), 1–54. https://jmlr.org/papers/v18/17-514.html
18.Ribeiro, M. T., Singh, S., & Guestrin, C. (2016). Why should I trust you? Explaining the predictions of any classifier. Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, 1135–1144. https://doi.org/10.1145/2939672.2939778
19.Rudin, C. (2019). Stop explaining black box machine learning models for high stakes decisions and use interpretable models instead. Nature Machine Intelligence, 1(5), 206–215. https://doi.org/10.1038/s42256-019-0048-x
20.Zhang, Y., Liu, D., Li, F., Zhao, Z., Liu, X., Gao, D., Zhang, Y., & Li, H. (2021). Identification of biomarkers for acute leukemia via machine learning-based stemness index. Gene, 804, 145903. https://doi.org/10.1016/j.gene.2021.145903
