In the recent decade, the importance of DNA damage repair (DDR) and its clinical application have been firmly recognized in prostate cancer (PC). For example, olaparib was just approved in May 2020 to treat metastatic castration-resistant PC with homologous recombination repair-mutated genes; however, not all patients can benefit from olaparib, and the treatment response depends on patient-specific mutations. This highlights the need to understand the detailed DDR biology further and develop DDR-based biomarkers. In this study, we establish a four-gene panel of which the expression is significantly associated with overall survival (OS) and progression-free survival (PFS) in PC patients from the TCGA-PRAD database. This panel includes DNTT, EXO1, NEIL3, and EME2 genes. Patients with higher expression of the four identified genes have significantly worse OS and PFS. This significance also exists in a multivariate Cox regression model adjusting for age, PSA, TNM stages, and Gleason scores. Moreover, the expression of the four-gene panel is highly correlated with aggressiveness based on well-known PAM50 and PCS subtyping classifiers. Using publicly available databases, we successfully validate the four-gene panel as having the potential to serve as a prognostic and predictive biomarker for PC specifically based on DDR biology.

The importance of anti-androgen therapy for prostate cancer (PC) has been well recognized. However, the mechanisms underlying prostate cancer resistance to anti-androgens are not completely understood. Therefore, identifying pharmacological targets in driving the development of castration-resistant PC is necessary. In the present study, we sought to identify core genes in regulating steroid hormone pathways and associating them with the disease progression of PC. The selection of steroid hormone-associated genes was identified from functional databases, including gene ontology, KEGG, and Reactome. The gene expression profiles and relevant clinical information of patients with PC were obtained from TCGA and used to examine the genes associated with steroid hormone. The machine-learning algorithm was performed for key feature selection and signature construction. With the integrative bioinformatics analysis, an eight-gene signature, including CA2, CYP2E1, HSD17B, SSTR3, SULT1E1, TUBB3, UCN, and UGT2B7 was established. Patients with higher expression of this gene signature had worse progression-free interval in both univariate and multivariate cox models adjusted for clinical variables. The expression of the gene signatures also showed the aggressiveness consistently in two external cohorts, PCS and PAM50. Our findings demonstrated a validated eight-gene signature could successfully predict PC prognosis and regulate the steroid hormone pathway.

Motivation: Lipids play crucial roles in various biological functions and diseases. However, a gap exists in databases providing information of lipids functions based on curated information. Consequently, LipidFun is purposed as the first lipid function database with sentence-level evidence detailing lipid-related phenotypes and biological functions. Results: Potential lipid functions were extracted from the biomedical literature using natural language processing techniques, with accuracy and reliability ensured through manual curation by four domain experts. LipidFun constructs classification systems for lipids, biological functions, and phenotypes for named entity recognition. Sentence-level evidence is extracted to highlight connections to lipid-associated biological processes and diseases. Integrating these classification systems and a large amount of sentence-level evidence allows LipidFun to provide an overview of lipid–phenotype and lipid–biological function associations through concise visualizations. Overall, LipidFun unravels the relationships between lipids and biological mechanisms, underscoring their overarching influence on physiological processes. Availability and implementation: LipidFun is available at https://lipidfun.bioinfomics.org/.

Background: Bladder cancer (BLCA) exhibits marked heterogeneity, and current classifiers provide limited guidance for prognosis or treatment. Because epigenetic reprogramming and metabolic rewiring jointly shape BLCA biology, we sought to identify epigenomically informed biomarkers with functional relevance. Methods: Epigenome (genome-wide promoter DNA methylation) and matched transcriptome (RNA sequencing) profiles from tumor and adjacent normal samples were integrated to identify genes with concordant differential methylation and expression patterns. A survival-oriented machine learning framework distilled these candidates into a 25-gene signature. The prognostic performance was evaluated in 4 independent BLCA cohorts. Multilayer characterization included the computational inference of tumor-infiltrating immune cells and in silico drug sensitivity prediction. The functional relevance of key lipid metabolic hub genes was confirmed by pharmacological inhibition in BLCA cell line models, followed by colony formation and migration assays. Results: The signature, enriched for cell cycle regulation and lipid metabolism, stratified patients into high- and low-risk groups across the discovery and 4 validation datasets. The prognostic value remained independent of age, pathological stage, and common genomic alterations. Low-risk tumors exhibited computationally inferred immune-inflamed phenotypes, whereas high-risk tumors exhibited lower immune engagement and lower half-maximal inhibitory concentration values for several drugs. Network analysis identified fatty acid synthase and stearoyl-coenzyme A desaturase as central nodes; their inhibition reduced BLCA cell proliferation and migration, supporting pathway-level functional relevance. Conclusion: By integrating epigenomic and transcriptomic layers with machine learning, we delineated a lipid-centric 25-gene signature that delivers stage-independent prognostication, illuminates tumor–immune interactions, and nominates actionable therapeutic targets. This experimentally vetted multi-omics framework advances precision oncology for BLCA.