Machine Learning-Based Classification of Malignant Glioblastoma Cells with Single-Cell RNA-Seq

Shengnan Liang

doi:10.54254/2753-8818/2026.AU29871

Theoretical and Natural ScienceOpen access

Machine Learning-Based Classification of Malignant Glioblastoma Cells with Single-Cell RNA-Seq

Research Article

Open Access

Machine Learning-Based Classification of Malignant Glioblastoma Cells with Single-Cell RNA-Seq

Shengnan Liang ^1*

¹ Huazhong University of Science and Technology

^*Corresponding author: u202212919@hust.edu.cn

Published on 26 November 2025

TNS Vol.152

ISSN (Print): 2753-8826

ISSN (Online): 2753-8818

ISBN (Print): 978-1-80590-565-3

ISBN (Online): 978-1-80590-566-0

Download Cover

Abstract

Glioblastoma is the most malignant primary brain tumor with high heterogeneity, making it challenging to achieve accurate diagnosis and evaluate treatment efficacy. With the fast development of single-cell RNA sequencing technology, malignant cells can be identified at the single-cell level to evaluate tumor purity. This study developed a computational workflow that integrated single-cell sequencing data and machine learning methods. Two classification models, XGBoost and a multilayer perceptron, were developed based on 30 selected genes with most differential expression identified by an independent samples t-tests from whole-genome expression data. Subsequently, the performance of two models was evaluated using multiple evaluation metrics. Experimental results showed that the two machine learning models had excellent performance in distinguishing malignant cells in glioblastoma. For distinguishing malignant cells, the AUC, accuracy, sensitivity and specificity of the XGBoost model were 0.941, 0.894, 0.883 and 0.901, respectively; while those of the MLP model were 0.937, 0.883, 0.865 and 0.896, respectively. In addition, the results of the probability distribution experiment showed that the XGBoost model had a more concentrated distribution, while the MLP model had a relatively broader distribution. These results were consistent with the effectiveness of the two machine learning approaches in malignant cell identification. This study validated the effectiveness of using machine learning methods based on single-cell RNA-seq data in identifying malignant cells in glioblastoma. This machine learning workflow could provide a reliable computational tool for subsequent malignant cell identification and tumor purity assessment.

Keywords:

Glioblastoma, Single-cell sequencing, MLP, XGBoost

View PDF

References

[1]. Mao, X. G. (2024). Understanding glioblastoma at the single-cell level: Latest progress and future challenges [J]. Chinese Journal of Neurosurgical Disease Research, 18(05), 21.

[2]. Zhao, H. H., Zhang, Z. Y., Yu, X. X., et al. (2025). Progress of single-cell RNA sequencing in cancer research [J]. Journal of Lanzhou University (Medical Sciences), 51(06), 79-86+94. https: //doi.org/10.13885/j.issn.2097-681X.2025.06.011.

[3]. Wang, J., & Su, W. (2022). Development history and research progress of related molecules in glioblastoma [J]. Journal of Practical Medical Techniques, 29(12), 1237-1240. https: //doi.org/10.19522/j.cnki.1671-5098.2022.12.001.

[4]. Sun, Y., Jin, W., Ni, H. B., et al. (2025). Research progress on the molecular mechanism of glucose metabolism in glioblastoma and its targeted therapy [J]. Journal of Clinical Neurosurgery, 22(05), 594-600.

[5]. Tian, L. Y., Cheng, C. D., Qian, Z. R., et al. (2024). Analysis of clinical and molecular pathological features and prognostic factors of primary glioblastoma [J]. Journal of Clinical Neurosurgery, 21(03), 292-297.

[6]. Zhang, H. B. (2024). A study on predicting TERT promoter mutation in glioblastoma based on multi-parameter MRI deep learning [D]. Southern Medical University. https: //doi.org/10.27003/d.cnki.gojyu.2024.000165.

[7]. Wang, G., Zhao, J. J., Nuerbiyaemu·Abulikemu, et al. (2022). Study on the effectiveness of subregional MRI radiomics combined with classification algorithms in predicting MGMT methylation in glioblastoma [J]. Journal of Clinical Radiology, 41(07), 1222-1226. https: //doi.org/10.13437/j.cnki.jcr.2022.07.034.

[8]. Lv, C., Xia, L. F., Chen, Q. S., et al. (2025). Predicting MGMT promoter methylation status in glioblastoma using deep neural network MRI radiomics [J]. Magnetic Resonance Imaging, 16(10), 35-40.

[9]. Li, Q., Dong, F., Jiang, B., et al. (2019). A study on the optimization of glioblastoma EGFR gene amplification classification model based on radiomics [C]// Zhejiang Medical Association, Radiology Branch of Zhejiang Medical Association. Proceedings of the 2019 Academic Congress of Radiology of Zhejiang Medical Association. The Second Affiliated Hospital of Zhejiang University School of Medicine, 2019: 34. https: //doi.org/10.26914/c.cnkihy.2019.101392.

[10]. Ren, L., Wei, Z. L., Qiu, Y. M., et al. (2015). Expression of ID1 in three glioblastoma cell lines and its significance [J]. Journal of Clinical Medicine in Practice, 19(09), 62-64.

[11]. Song, P., Deng, H., & Zhang, M. X. (2024). Application of single-cell RNA sequencing technology in glioblastoma research [J]. Journal of Huazhong University of Science and Technology (Medical Sciences), 53(01), 110-116.

[12]. Deng, H., Guo, W. J., Song, P., et al. (2023). Construction of a prognostic model for glioblastoma based on single-cell sequencing [J]. Biotechnology, 33(01), 38-47. https: //doi.org/10.16519/j.cnki.1004-311x.2023.01.0007.

[13]. Couturier, C.P., Ayyadhury, S., Le, P.U. et al. Single-cell RNA-seq reveals that glioblastoma recapitulates a normal neurodevelopmental hierarchy. Nat Commun 11, 3406 (2020). https: //doi.org/10.1038/s41467-020-17186-5

[14]. Zhang, Y. Z. (2024). A study on differentiating glioblastoma from solitary brain metastases based on radiomics and deep learning [D]. Southern Medical University. https: //doi.org/10.27003/d.cnki.gojyu.2024.000735.

[15]. an, J., Sun, Q., Tan, X., et al. (2023). Imaging-based deep learning identifies glioblastoma risk stratification with genomic and transcriptomic heterogeneity: A multicenter study [J]. International Journal of Medical Radiology, 46(02), 244-245. https: //doi.org/10.19300/j.2023.e0204.

[16]. Zheng, F., & Chen, X. Z. (2022). Research progress of deep learning in glioblastoma [J]. Magnetic Resonance Imaging, 13(03), 115-117.

References

[1]. Mao, X. G. (2024). Understanding glioblastoma at the single-cell level: Latest progress and future challenges [J]. Chinese Journal of Neurosurgical Disease Research, 18(05), 21.

[10]. Ren, L., Wei, Z. L., Qiu, Y. M., et al. (2015). Expression of ID1 in three glioblastoma cell lines and its significance [J]. Journal of Clinical Medicine in Practice, 19(09), 62-64.

[16]. Zheng, F., & Chen, X. Z. (2022). Research progress of deep learning in glioblastoma [J]. Magnetic Resonance Imaging, 13(03), 115-117.

Cite this article

Liang,S. (2025). Machine Learning-Based Classification of Malignant Glioblastoma Cells with Single-Cell RNA-Seq. Theoretical and Natural Science,152,1-10.

Data availability

The datasets used and/or analyzed during the current study will be available from the authors upon reasonable request.

About volume

Volume title: Proceedings of ICMMGH 2026 Symposium: Biomedical Imaging and AI Applications in Neurorehabilitation

ISBN: 978-1-80590-565-3(Print) / 978-1-80590-566-0(Online)

Editor: Sheiladevi Sukumaran, Alan Wang

Conference website: https://www.icmmgh.org/auckland.html

Conference date: 14 November 2025

Series: Theoretical and Natural Science

Volume number: Vol.152

ISSN: 2753-8818(Print) / 2753-8826(Online)

© 2024 by the author(s). Licensee EWA Publishing, Oxford, UK. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license. Authors who publish this series agree to the following terms:

1. Authors retain copyright and grant the series right of first publication with the work simultaneously licensed under a Creative Commons Attribution License that allows others to share the work with an acknowledgment of the work's authorship and initial publication in this series.

2. Authors are able to enter into separate, additional contractual arrangements for the non-exclusive distribution of the series's published version of the work (e.g., post it to an institutional repository or publish it in a book), with an acknowledgment of its initial publication in this series.

3. Authors are permitted and encouraged to post their work online (e.g., in institutional repositories or on their website) prior to and during the submission process, as it can lead to productive exchanges, as well as earlier and greater citation of published work (See Open access policy for details).