Computational Biology and Chemistry

cervical nuclei segmentation through synergic conditional generative adversarial network in cervical smear images

Cervical nuclei segmentation is critical for the early detection and accurate diagnosis of cervical cancer. However, this task is challenging due to the presence of clumped nuclei and variations in texture, shape, and contrast. To address these challenges, we proposed a novel synergic conditional generative adversarial network (SCGAN) for cervical nuclei segmentation. The SCGAN integrates densely connected blocks that progressively extract hierarchical features, a Unified Attention Module (UAM) for selective feature refinement and the Scale-Adaptive Feature Integration and upsampling (SAFIU) module for multi-scale feature integration and upsampling, and a synergic discriminator to enhance adversarial learning. The SAFIU module constructs a multi-scale feature pyramid by progressively upsampling across feature levels, effectively retaining fine spatial details critical for segmenting small nuclei. The Scale-Adaptive Fusion (SAF) block further facilitates feature learning by merging high-level features with low-level spatial cues from the encoder, and then forwarding the fused representation to the corresponding decoder stage. On the adversarial side, the synergic discriminator, consisting of ResNet-50 and EfficientNet-B2, is designed for collaborative learning and accelerates convergence with the help of a synergic block. The integration of an Uncertainty-Aware Attention (UAA) mechanism in the synergic block helps the discriminators concentrate on ambiguous or overlapping regions, thereby providing more informative feedback to the generator. Experiments on multiple cervical nuclei datasets demonstrated that the proposed SCGAN outperformed existing methods in terms of sensitivity, specificity, Dice coefficient, and F1-score. By effectively integrating multi-scale features and leveraging adversarial training, our SCGAN achieves more accurate and more consistent cervical nuclei segmentation, paving the way for improved computer-aided diagnosis systems.

Identification and validation of miR-21 key genes in cervical cancer through an integrated bioinformatics approach

Cervical cancer is one of the most prevalent female reproductive cancers. miR-21 is a multi-target oncomiR that has shown its potential in regulating several cancers including colon, pancreatic, breast, prostate, ovarian, and cervical cancer. However, the signaling network of miR-21 remains underexplored, and only a limited number of miR-21 gene targets in cervical cancer have been reported. In this context, the present study was undertaken to evaluate the role of miR-21 in cervical cancer by combining in silico analysis with in vitro validation in cervical cancer cells. The miR-21 target genes were predicted using four different prediction tools: miRWalk, DIANA, miRDB, and TargetScan. A total of 113 overlapping target genes, common in at least three of the prediction tools, were shortlisted and subjected to functional enrichment analysis. The analysis predicted that JAK-STAT, MAPK, neurotrophin, and Ras signaling pathways are significantly (p≤0.05) targeted by miR-21. The MCODE plugin identified the potential cluster in the protein-protein interaction network based on the highest degree of connectivity. After GEPIA2 validation of all 20 hub genes, NTF3, LIFR, and IL-6R were shortlisted for validation in cervical cancer cell lines. The results showed that NTF3, LIFR, and IL-6R were significantly upregulated in the miR-21 knockdown CaSki cell lines in 6.27, 1.92 and 1.71 folds (p≤0.01), respectively. Similarly, in HeLa cell lines expression of NTF3, LIFR, and IL-6R were overexpressed in 4.06, 5.65, 2.42 folds (p≤0.001), respectively. Findings of the study was confirming the role of miR-21 in regulating the expression of these genes. Additionally, the knockdown of miR-21 significantly inhibited the secretion of matrix metalloproteinases by CaSki cells. These results highlight that miR-21 could be a potential therapeutic target for cervical cancer, although further preclinical and clinical studies are required to validate its role and efficacy.

MMDAE-HGSOC: A novel method for high-grade serous ovarian cancer molecular subtypes classification based on multi-modal deep autoencoder

High-grade serous ovarian cancer (HGSOC) is a type of ovarian cancer developed from serous tubal intraepithelial carcinoma. The intrinsic differences among molecular subtypes are closely associated with prognosis and pathological characteristics. At present, multi-omics data integration methods include early integration and late integration. Most existing HGSOC molecular subtypes classification methods are based on the early integration of multi-omics data. The mutual interference among multi-omics data is ignored, which affects the effectiveness of feature learning. High-dimensional multi-omics data contains genes unassociated with HGSOC molecular subtypes, resulting in redundant information, which is not conducive to model training. In this paper, we propose a multi-modal deep autoencoder learning method, MMDAE-HGSOC. MiRNA expression, DNA methylation, and copy number variation (CNV) are integrated with mRNA expression data to construct a multi-omics feature space. The multi-modal deep autoencoder network is used to learn the high-level feature representation of multi-omics data. The superposition LASSO (S-LASSO) regression algorithm is proposed to fully obtain the associated genes of HGSOC molecular subtypes. The experimental results show that MMDAE-HGSOC is superior to the existing classification methods. Finally, we analyze the enrichment gene ontology (GO) terms and biological pathways of these significant genes, which are discovered during the gene selection process.

Robust biomarker screening from gene expression data by stable machine learning-recursive feature elimination methods

Recently, identifying robust biomarkers or signatures from gene expression profiling data has attracted much attention in computational biomedicine. The successful discovery of biomarkers for complex diseases such as spontaneous preterm birth (SPTB) and high-grade serous ovarian cancer (HGSOC) will be beneficial to reduce the risk of preterm birth and ovarian cancer among women for early detection and intervention. In this paper, we propose a stable machine learning-recursive feature elimination (StabML-RFE for short) strategy for screening robust biomarkers from high-throughput gene expression data. We employ eight popular machine learning methods, namely AdaBoost (AB), Decision Tree (DT), Gradient Boosted Decision Trees (GBDT), Naive Bayes (NB), Neural Network (NNET), Random Forest (RF), Support Vector Machine (SVM) and XGBoost (XGB), to train on all feature genes of training data, apply recursive feature elimination (RFE) to remove the least important features sequentially, and obtain eight gene subsets with feature importance ranking. Then we select the top-ranking features in each ranked subset as the optimal feature subset. We establish a stability metric aggregated with classification performance on test data to assess the robustness of the eight different feature selection techniques. Finally, StabML-RFE chooses the high-frequent features in the subsets of the combination with maximum stability value as robust biomarkers. Particularly, we verify the screened biomarkers not only via internal validation, functional enrichment analysis and literature check, but also via external validation on two real-world SPTB and HGSOC datasets respectively. Obviously, the proposed StabML-RFE biomarker discovery pipeline easily serves as a model for identifying diagnostic biomarkers for other complex diseases from omics data. The source code and data can be found at https://github.com/zpliulab/StabML-RFE.

Unraveling the biological functions of UCEC: Insights from a prognostic signature model

Uterine corpus endometrial carcinoma (UCEC) is a prevalent gynecological tumor with a bleak prognosis. Anomalous glycosylation plays a pivotal role in tumorigenesis. Currently, there is a lack of prognostic signatures based on glycosylation-related genes for UCEC. Thus, our research aims to construct a predictive model and validate the correlation between relevant genes and biological functions. Using the TCGA database, we developed prognostic models and explored their relationships with survival outcomes. We further selected key genes to verify their expression in tissues and assess their impact on cellular behavior. The clinical prognosis of the high-risk group was significantly worse than that of the low-risk group. The nomogram model accurately predicted UCEC patient prognosis. Additionally, we identified OLFML1 as a unique signature gene that can inhibit UCEC progression and reduce radiation resistance in vitro. Our model, which is based on glycosylation-related genes in UCEC, effectively identifies high-risk patients and provides valuable prognostic information. In addition, OLFML1 acts as a tumor suppressor in UCEC and enhances radiosensitivity, suggesting a new potential target for improving therapeutic efficacy.

Computational-based identification and analysis of globally expressed differential genes in high-grade serous ovarian carcinoma cell lines

Ovarian Cancer (OVCA) is the most occurring gynecological cancer worldwide, often diagnosed at a later stage and ultimate results in a high death rate. To overcome this serious health concern, it is important to understand the molecular mechanisms and equally significant to identify the putative biomarkers as well as the therapeutic drug targets for the early diagnosis and treatment of OVCA. In doing so, a strategy is designed to study the most frequently diagnosed cases of OVCA called as High-Grade Serous Ovarian Carcinoma (HGSOC) cell lines with the combination of computational biology, biostatistics and cancer informatics approaches. This study is directed to investigate the global gene expression profiling, and to perform the analyses of identified global Differently Expressed Genes (DEGs) of OVCA. The microarray dataset (GSE71524) is comprised of tumor and cell line samples of OVCA and it was used for the identification of DEGs in the current study. The STRING database was used to construct Protein-Protein Interaction (PPI) network of DEGs, and hub genes were identified by the CytoHubba. In addition, a functional enrichment analysis of up- and down-regulated DEGs was performed by a bioinformatics database called as DAVID. The microRNAs (miRNAs) and transcription factors (TFs) analyses were conducted with the aid of biological tools, MAGIA and GenCOdis3, respectively. As a result, the genes comprised of CSF1R, TYROBP, PLEK, FGR, ACLY, ACACA, LAPTM5, C1 or f162, IL10RA and CD163 were identified as hub genes. Additionally, miRNA analysis resulted in finding an association of zinc finger protein with OVCA comes out after implementing different algorithms. On the other hand, in the TFs analysis resulted in various DEGs that were enriched by NFAT, NF1 and GABP TFs. In this study, it was observed that ACACA, ACLY and CSF1R DEGs showed significant occurrence in different steps, and therefore, these genes were studied, precisely. Nevertheless, the results may help to discover the potential biomarkers with deep understanding of molecular mechanisms. However, further validation is required to explain the OVCA pathogenesis.

Identification of key classification features of early cervical squamous cell carcinoma

Despite the tremendous progress in molecular analysis of pan-cancer, little is known regarding molecular classification of cervical squamous cell carcinoma. In this study, we adopted a multi-omics approach to identify potential key classification features of cervical squamous cell carcinoma. Specifically, we analyzed mRNA, and microRNA (miRNA) expression data, as well as DNA methylation and copy number variation in cervical squamous cell carcinoma cases, using datasets obtained from The Cancer Genome Atlas (TCGA). Moreover, we identified molecules in each dimension, as well as integrated and clustered filtered classification features, and used them to distinguish different subtypes. The resulting key classification features were used to establish a classification model for cervical squamous cell carcinoma. Our results revealed two cervical squamous cell carcinoma subtypes, with significant differences across clinical survival levels, as well as 8 key classification features of cervical squamous cell carcinomas. These findings are expected to provide important references for early classification of cervical squamous cell carcinoma and identification of classification markers.

Chinese herbal extract Astragalus radix potentiates human ovarian cancer cell cytotoxicity by aggravated ROS production and apoptosis

Ovarian cancer remains one of the most lethal gynaecological malignancies due to its late diagnosis, and resistance to conventional therapies. Traditional Chinese Medicine (TCM) is increasingly explored for its potential in cancer treatment. This study investigates the anti-tumor effects of a Chinese herbal extract on an ovarian cancer cell line in vitro. The ovarian cancer cell lines OVCAR-3 and SK-OV-3 treated with varying concentrations of the Chinese herbal extract (Astragalus radix) at different course of time. Cell viability using the MTT assay, and apoptosis was examined by flow cytometry after staining with Annexin V/PI staining. Molecular docking and dynamics were carried out to examine the interaction of quinacetol with a well-known target of ovarian cancer, i.e., phosphoinositide 3-kinase (PI3K). The Chinese herbal extract Astragalus radix significantly reduced the viability of ovarian cancer cells in a time- and dose- dependent way. Flow cytometry analysis revealed increased apoptotic rates in ovarian cancer cells compared to controls. Quinacetol was found to interact at active site of PI3K with binding energy of -6.9 kcal/mol. The PI3K-quinacetol complex was stable at physiological conditions as evident from molecular simulation studies. The findings of this study demonstrate that the Chinese herbal extract (Astragalus radix) exhibits potent anti-tumor effects against ovarian cancer cells in vitro, highlighting its potential as an adjunct or alternative therapeutic option. Further in vivo studies in animal models and clinical trials are warranted to explore the efficacy and safety of this herbal treatment in ovarian cancer patients.

Ovarian cancer detection from mutual information-ranked clinical biomarkers using an explainable attention-based residual multilayer perceptron

Ovarian cancer is a major health concern for women, contributing to substantial mortality and morbidity. Timely identification of ovarian cancer is crucial for enhancing patient health and survival rates. Current diagnostic practices involve the manual analysis of various clinical biomarkers to detect ovarian cancer. However, this approach can be subjective, time-consuming, and dependent on the expertise of the medical professional. To optimize workflow efficiency and improve diagnosis accuracy, we develop an automated deep learning model, called EA-ResMLP, which integrates a residual multilayer perceptron with squeeze-and-excitation attention block and explainable artificial intelligence. The integration of residual connections and attention mechanisms contributes to improved diagnostic accuracy by enabling deeper feature learning and emphasizing the most informative features through adaptive recalibration. The experimental results demonstrated that proposed method achieved an accuracy of 92.05%, indicating a 7.98% improvement over the conventional multilayer perceptron. Furthermore, the predictions of the EA-ResMLP model are analyzed using explainable artificial intelligence techniques such as local interpretable model-agnostic explanations, which generate feature contribution charts to highlight the impact of each input feature on the prediction. By integrating model predictions with feature contribution charts, the proposed model provides an explainable framework for ovarian cancer detection.

Computational insights into irinotecan's interaction with UBE2I in ovarian and endometrial cancers

Endometrial and Ovarian cancers are two highly prevalent and fatal reproductive diseases with poor prognoses among women. Elevated estrogen levels in Ovarian Cancer (OC) stimulate the endometrium, causing Endometrial Cancer (EC). Although numerous studies have reported the crucial genes and pathways in this cancer, the pathogenesis of this disease remains unclear. In this study, used bioinformatics tools to analyse GSE63678, GSE115810, GSE36389, GSE26712, GSE36668, GSE27651, GSE6008, GSE69429, GSE69428, GSE18521, GSE185209, GSE54388 gene expression microarray datasets for both the cancers. We analyzed the differential gene expression, functional association, and structural studies. The analysis identified crucial differentially expressed genes (DEGs) in both cancers associated with DNA damage, DNA integrity, and cell-cycle checkpoint signaling pathways. CLDN7, UBE2I, WT1, JAM2, FOXL2, F11R, JAM3, ZFPM2, MEF2C, and PIAS1 are the top 10 hub genes commonly identified in both cancer types. Only CLDN7 and F11R are upregulated, whereas the remaining hub genes are downregulated in both cancers, suggesting a common framework for contributing to tumorigenesis. Molecular docking and dynamics were performed on the UBE2I protein with Irinotecan Hydrochloride, which could serve as the new approach for treating and managing both cancers. The study reveals the common molecular pathways, pointing out the role of cell cycle and DNA damage and integrity checkpoint signaling in the pathogenesis of both cancer types. This study explored the UBE2I gene as a potential biomarker in OC and EC. Further, this study concludes that the irinotecan hydrochloride drug has higher therapeutic effects on UBE2I protein through docking and dynamics studies.

TransNeT-CGP: A cluster-based comorbid gene prioritization by integrating transcriptomics and network-topological features

The local disruptions caused by the genes of one disease can influence the pathways associated with the other diseases resulting in comorbidity. For gene therapies, it is necessary to prioritize the key genes that regulate common biological mechanisms to tackle the issues caused by overlapping diseases. This work proposes a clustering-based computational approach for prioritising the comorbid genes within the overlapping disease modules by analyzing Protein-Protein Interaction networks. For this, a sub-network with gene interactions of the disease pair was extracted from the interactome. The edge weights are assigned by combining the pairwise gene expression correlation and betweenness centrality scores. Further, a weighted graph clustering algorithm is applied and dominant nodes of high-density clusters are ranked based on clustering coefficients and neighborhood connectivity. Case studies based on neurodegenerative diseases such as Amyotrophic Lateral Sclerosis- Spinal Muscular Atrophy (ALS-SMA) pair and cancers such as Ovarian Carcinoma-Invasive Ductal Breast Carcinoma (OC-IDBC) pair were conducted to examine the efficacy of the proposed method. To identify the mechanistic role of top-ranked genes, we used Functional and Pathway enrichment analysis, connectivity analysis with leave-one-out (LOO) method, analysis of associated disease-related protein complexes, and prioritization tools such as TOPPGENE and Heml2.0. From pathway analysis, it was observed that the top 10 genes obtained using the proposed method were associated with 10 pathways in ALS-SMA comorbidity and 15 in the case of OC-IDBC, while that in similar methods like SAPDSB and S2B were 4, 6 respectively for ALS-SMA and 9, 10 respectively for OC-IDBC. In both case studies, 70 % of the disease-specific benchmark protein complexes were linked to top-ranked genes of the proposed method while that of SAPDSB and S2B were 55 % and 60 % respectively. Additionally, it was found that the removal of the top 10 genes disconnect the network into 14 distinct components in the case of ALS-SMA and 9 in the case of OC-IDBC. The experimental results shows that the proposed method can be effectively used for identifying key genes in comorbidity and can offer insights about the intricate molecular relationship driving comorbid diseases.

GC-PGE: A novel deep learning model for tumor drug resistance prediction and core resistance gene extraction based on graph and signaling pathways

PTTG1 as a common promising target for PCOS, Ovarian Cancer, and Major Depressive Disorder patients

Women are susceptible to hormonal imbalances and endocrine-related disorders such as Polycystic Ovary Syndrome (PCOS), Ovarian Cancer (OC), and Major Depressive Disorder (MDD). This study aims to identify gene-level interconnections among these conditions using omics-based bioinformatic approaches. Publicly available GEO datasets, viz., GSE226146 (PCOS), GSE18520 (OC), and GSE125664 (MDD), were analyzed, which in total resulted in 21,366 differentially expressed genes (DEGs), including 11,174 upregulated and 10,198 downregulated genes. Common genes PTTG1 and PID1 were identified using Venny 2.0. A protein-protein interaction (PPI) network was constructed using STRING, and 10 hub genes (ANAPC5, ANAPC2, PTTG1, FZR1, ANAPC4, CDC20, CDC27, ANAPC10, UBE2C, and BUB1) were identified using CytoHubba based on MCC scoring. Functional enrichment analysis showed significant involvement of these genes in oocyte meiosis, progesterone-mediated oocyte maturation, mitotic regulation, and metaphase-anaphase transition (p < 0.05). PTTG1, identified as both a common and hub gene, was downregulated in PCOS and upregulated in OC and MDD. Drug-gene interaction analysis using DSigDB via Enrichr identified Alvespimycin (for PCOS) and Gefitinib (for OC) as drugs targeting PTTG1. Molecular docking using AutoDock 4.2.6 showed that Alvespimycin and Ephedrone bind PTTG1 with a binding affinity of - 4.59 kcal/mol and - 5.81 kcal/mol, respectively, while Gefitinib showed - 4.92 kcal/mol, slightly less than Troglitazone (-5.3 kcal/mol) for OC. This study highlights PTTG1 as a shared molecular link among PCOS, OC, and MDD, suggesting its potential as a therapeutic target and providing insights into the genetic and physiological overlap of these conditions.

MethylMSI: Prediction of microsatellite instability based on DNA methylation profile and SVM model

Microsatellite instability (MSI) has gained increasing attention as a promising biomarker for cancer immunotherapy in diverse cancer types. As a result of DNA mismatch repair (MMR) deficiency, MSI tumors show a hypermutator phenotype. Based on this feature, many computational methods have been developed to predict MSI with DNA sequencing data or mRNA expression data. Using the methylation data from The Cancer Genome Atlas (TCGA) program, we developed MethylMSI, a support vector machine (SVM) model based tool for MSI classification in three cancer cohorts (colorectal, stomach and endometrial cancers). We demonstrated that MethylMSI has high MSI prediction accuracy in all three cohorts. Analysis of underlying features of the SVM model showed that hypermethylation of MLH1 is an important indicator for MSI in colorectal and endometrial cancers; while in stomach cancer, other MMR deficiency might be more important for MSI formation. We concluded that SVM had the potential to be the best model for predicting MSI status from DNA methylation data. In addition, an easy handling software is now freely available at https://github.com/Huanglab-ai/MethylMSI.

Identification of hub genes as potential diagnostic biomarkers for cervical cancer: A bioinformatic approach

Cervical cancer remains a prevalent malignancy with rising incidence, primarily due to sexual transmission, persistent HPV infection, and delayed screening. Identifying new biomarkers is critical for improved diagnosis, prognosis, and treatment of cervical cancer. This study utilized integrated bioinformatics to identify potential biomarkers by analysing gene expression data from the GEO database. Four GEO microarray datasets (GSE7410, GSE7803, GSE52903, GSE67522) were analysed using GEO2R to identify DEGs with an adjusted p-value <0.05. Common DEGs were visualized using Venn diagrams. Protein-protein interaction network was constructed using STRING to identify hub genes. Gene Ontology (GO) and KEGG pathway analyses were performed to investigate biological functions and pathways. The Human Protein Atlas (HPA) was used for in silico validation of protein expression via immunohistochemistry. Kaplan-Meier survival analysis was performed to determine the prognostic value of hub genes. Analysis revealed 684 common DEGs across the datasets (446 upregulated, 238 downregulated). The top 20 upregulated DEGs from GSE67522 were used for heatmap construction and PPI analysis, leading to the identification of nine key hub genes. GO and KEGG analyses showed that six of these were significantly involved in cell cycle regulation and tumorigenic pathways. These hub genes were validated for their protein expression through HPA data. Six hub genes (CCNB2, AURKA, CDC20, CDT1, CENPF, and KIF2C) were identified as potential biomarkers for cervical cancer management. These findings provide valuable insight into the molecular profiles of genes that play significant roles in cervical cancer for translational outcomes in diagnosis.

Reliable estrogen-related prognostic signature for uterine corpus endometrial carcinoma

Uterine corpus endometrial carcinoma (UCEC) is a predominant gynecological malignancy worldwide. Overdosed estrogen exposure has been widely known as a crucial risk factor for UCEC patients. The purpose of this work is to explore crucial estrogen-related genes (ERGs) in UCEC. UCEC scRNA-seq data, bulk RNA data, and ERGs were obtained from GEO, TCGA, and Molecular Signature Database, respectively. Differential expression analysis and cross analysis determined the candidate genes, and optimal genes in risk score were obtained after univariate Cox regression analysis, LASSO Cox regression analysis, and multivariate Cox regression analysis. The functional information was revealed by GO, KEGG, and GSVA enrichment analyses. CCK8 assay was used to detect the drug sensitivity. After cross analysis of the differentially expressed genes and the 8734 ERGs, 86 differentially expressed ERGs were identified in UCEC, which were significantly enriched in some immune related pathways and microbiota related pathways. Of them, the most optimal 8 ERGs were obtained to build prognostic risk score, including GAL, PHGDH, SLC7A2, HNMT, CLU, AREG, MACC1, and HMGA1. The risk score could reliably predict patient prognosis, and high-risk patients had worse prognosis. Higher HMGA1 gene expression exhibited higher sensitivity to Osimertinib. Predictive risk score based on 8 ERGs exhibited excellent prognostic value in UCEC patients, and high-risk patients had inferior survival. UCEC patients with distinct prognoses showed different tumor immune microenvironment.

R-loop-driven molecular subtypes reveal prognostic and immunogenomic features in uterine corpus endometrial carcinoma

R-loops are three-stranded nucleic acid structures implicated in genome instability and cancer progression. However, the prognostic significance and mechanistic role of R-loops in uterine corpus endometrial carcinoma (UCEC) remain poorly understood. Transcriptomic, clinical, mutational, and spatial data for UCEC were obtained from The Cancer Genome Atlas (TCGA) and public databases. Multiomics analyses, including prognostic modeling, survival analyses, differential expression analyses, copy number variation (CNV) profiling, somatic mutation comparisons, single-cell transcriptomics, spatial transcriptomics, and immune-related pathway exploration, were conducted to elucidate the biological implications of R-loop genes, matrix-specific CSDE1, and the associated SPP1 pathway. In vivo and in vitro functional experiments were conducted to evaluate the role of CSDE1 in UCEC. Elevated R-loop activity was associated with advanced clinical stage, high tumor grade, and poor survival outcomes in patients with UCEC. A robust prognostic model based on R-loop genes achieved high predictive accuracy across multiple datasets. Low-risk patients had higher tumor mutation burdens and distinct mutational profiles, whereas high-risk patients had more chromosomal instability and more CNV events. CSDE1 emerged as the top predictive gene, displaying fibroblast-specific expression and copy number-driven upregulation. Single-cell and spatial transcriptomics revealed that CSDE1⁺ fibroblasts actively communicated with immune cells via the SPP1 pathway and were spatially enriched in malignant, fibroblast-dense regions. High CSDE1 expression correlated with the activation of oncogenic pathways and the suppression of multiple steps in the cancer-immunity cycle. Furthermore, CSDE1 promoted the proliferation and migration of UCEC cells in vitro and in vivo by reducing R-loop accumulation and DNA damage. R-loop activity and CSDE1 expression define a clinically relevant molecular program in UCEC that integrates genomic instability, immunosuppression, and stromal remodeling. These findings provide a basis for stratified prognosis and potential therapeutic targeting in endometrial cancer, suggesting that CSDE1 may be a promising new therapeutic target for the treatment of UCEC in the future.