bioRxiv Science⌕ Search
Home/Archive/bioRxiv

bioRxiv · 10.1101/2025.02.28.640917

Integrative Machine Learning Reveals Potential Signature Genes Using Transcriptomics in Colon Cancer

Hamza, M. A.·Islam, M. S.
View original source ↗Export metadata ↓

Abstract

BackgroundColon cancer is a significant health burden in the world and the second leading cause of cancer-related deaths. Despite advancements in diagnosis and treatment, identifying robust biomarkers for early detection and therapeutic targets remains imperative. Materials and methodsThis study used an integrative approach combining transcriptomics and machine learning to identify signature genes and pathways associated with colon cancer. RNA-Seq data from The Cancer Genome Atlas-Colon Adenocarcinoma (TCGA-COAD) project, comprising 485 samples (444 tumors and 41 normal tissues), were analyzed. ResultsDifferential gene expression analysis revealed 657 upregulated and 8,566 downregulated genes. Notably, EPB41L3, TSPAN7, and ABI3BP were identified as highly upregulated, while LYVE1, PLPP1, and NFE2L3 were significantly downregulated in tumor samples. Gene Set Enrichment Analysis (GSEA) identified dysregulated pathways, including E2F targets, MYC targets, and G2M checkpoints, underscoring cell cycle regulation and metabolic reprogramming alterations in colon cancer. Machine learning models-Random Forest, Neural Networks, and Logistic Regression-achieved high classification accuracy (97-99%) and near-perfect Area Under the Receiver Operating Characteristic Curve (AUC-ROC) values (approximately 1.00), validating their predictive capabilities. Key genes consistently identified across these models highlight their potential translational relevance as biomarkers. This study integrates differential expression analysis, pathway enrichment, and machine learning to uncover critical insights into colon cancer biology. ConclusionThe findings lay the groundwork for developing diagnostic and therapeutic strategies, with the identified genes and pathways serving as promising candidates for future validation and clinical applications. This approach exemplifies the potential of precision medicine to advance colon cancer research and improve patient outcomes.

Source connections

Explore related subjects

cancer biologyPreprint

Keep this discovery

Explore connections, maps & timelines

BibTeXRIS

Hamza, M. A., Islam, M. S.. 2025-03-06. Integrative Machine Learning Reveals Potential Signature Genes Using Transcriptomics in Colon Cancer. https://doi.org/10.1101/2025.02.28.640917

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related preprints

bioRxiv2026

Ex vivo human tumor slices more accurately predict patient responses to an oncolytic virus than in vivo mouse models

Immunotherapies, including oncolytic viruses (OV), are promising therapies that can enhance anti-tumor immune responses. However, preclinical success of immunotherapies in mouse models has not always translated to clinical benefit in cancer patients. This study compared preclinical efficacy and mechanism of action for ASP9801, a vaccinia virus expressing IL-7 and IL-12, using mouse models of colorectal cancer (CRC) in vivo and in human organotypic tumor slice models ex vivo. The murine surrogate for ASP9801 significantly reduced tumor volumes in treated and abscopal tumors in two different CRC models in vivo (MC38 and RO100). Treatment efficacy was accentuated when combined with anti-PD1 treatment, and single-cell RNA sequencing analysis revealed depletion of tumor cells and increased T cell infiltration and activation in both treated and abscopal tumors. However, human tissue analysis ex vivo (E-slices) using PDX models and patient samples showed that ASP9801 is not effective in CRC, consistent with clinical trial results. On the other hand, ASP9801 was highly effective in GBM, indicating indication-specific efficacy of ASP9801, and how E-slice assays can be used to identify treatment-sensitive indications. This study demonstrates the superiority of E-slices over mouse models for predicting clinical response and its utility in planning clinical trials.

cancer biology↗
bioRxiv2026

Immune-cell depleted diffuse large B-cell lymphomas have reduced expression of MHC class I

Immunotherapy has transformed treatment for many cancers. In the aggressive and genetically heterogeneous diffuse large B-cell lymphoma (DLBCL), CD19 CAR T-cell therapy is highly effective, whereas immune checkpoint blockade has shown limited benefit. Loss of MHC expression is a common mechanism to escape T-cell cytotoxicity, and loss of MHC class I (MHC-I) and II are frequent in DLBCL. We applied imaging mass cytometry to diagnostic biopsies from younger, high-risk DLBCL patients to map the tumor microenvironment (TME) spatial architecture in relation to tumor cell MHC expression, mutational status, transcriptomic and proteomic profiles. Neighborhood analyses identified four TME subtypes: immune-cell depleted and three immune-infiltrated types (mixed, CD4 T cell-rich, CD8 T-cell/macrophage-rich). Depleted cases had shorter overall survival (p = 0.033) and increased expression of proteins involved in DNA replication and proliferation markers compared to infiltrated cases. Tumor cell MHC-I expression was heterogeneous. Cases with low frequency of MHC-I-pos tumor cells were enriched for the depleted TME type. MHC-I-pos tumor cells were surrounded by CD4 and CD8 T cells and M1 macrophages, whereas MHC-I-neg tumor cells were closer to other MHC-I-neg tumor cells. These findings suggest that TME-based classification incorporating tumor cell MHC-I status may improve individualized immunotherapy selection.

cancer biology↗
bioRxiv2026

Cross-species analysis links cell-cell communication rewiring to NOTCH2 during serous endometrial carcinogenesis

Cell-cell interactions shape the fate of mutant cells during cancer initiation but how these interactions evolve during progression to pathologically recognizable lesions remain poorly understood. Here, we investigated cell-cell communication during serous endometrial carcinoma (SEC; also known as uterine serous carcinoma) development using a lineage-traceable mouse model and cross-species analyses of the mouse and human neoplastic endometrium. In mice, the early, pre-dysplastic stage was marked by a global decrease in inferred cell-cell interactions, followed by extensive communication network rewiring during neoplastic progression. Pathway-specific analysis revealed a similar pattern for NOTCH signaling, with NOTCH2 emerging as the dominant NOTCH receptor in Trp53/Rb1-mutant immature epithelial cells. Functionally, NOTCH2 promoted the outgrowth of more proliferative mutant organoids. Cross-species transcriptomic analysis identified conserved immature epithelial states in mouse and human neoplastic endometrial epithelium. In human tissues, NOTCH2 was overexpressed in serous endometrial intraepithelial carcinoma, a precursor of SEC, and in overt SEC. Furthermore, elevated NOTCH2 expression was associated with poor patient survival. These findings link cell-cell communication rewiring during experimental SEC development to conserved neoplastic epithelial states and identify NOTCH2 as an early marker and a potential target of disease interception.

cancer biology↗