Integrating the advantages of machine learning with the rapidly accumulating high-throughput sequencing data facilitates our capacity for biological discovery and the advancement of molecular medicine. In recent years, bulk RNA-seq technology has established itself as a cost-effective and widely used method for obtaining complete transcriptome profiles of test samples, enabling the identification of key cancer-associated expression patterns. Various machine learning algorithms, in turn, enable the development of informative diagnostic and prognostic models, ensuring the efficient processing of high-dimensional RNA-Seq data. The convergence of these methods shows great promise for oncology. In this narrative review, we describe bulk RNA-Seq-based ML models in oncology as a complete workflow from data preprocessing to model validation. We provide practical recommendations for algorithm selection and study design, and discuss bulk RNA-Seq deconvolution as a cost-effective alternative to single-cell RNA-Seq for analyzing tumor cellular composition. These insights offer a practical guide for developing reproducible diagnostic and prognostic models with translational potential.

Paragangliomas and pheochromocytomas (PPGLs) are rare neuroendocrine tumors originating from paraganglionic tissue in many sites of the body. Most PPGLs are characterized by nonaggressive behavior but all of them have the potential to metastasize. PPGLs represent a great diagnostic dilemma as it is difficult to recognize tumors that are likely to be metastasizing; criteria of malignancy can be found both in benign and metastatic forms. This review aims to analyze the current knowledge of the nature of metastasizing PPGLs paying particular attention to head and neck paragangliomas (HNPGLs). Potential predictors of the malignancy risk for PPGLs were summarized and discussed. These data may also help in the development of diagnostic and prognostic strategies, as well as in the identification of novel potential therapeutic targets for patients with PPGLs.

Carotid paragangliomas (CPGLs) are rare neuroendocrine tumors often associated with mutations in SDHx genes. The immunohistochemistry of succinate dehydrogenase (SDH) subunits has been considered a useful instrument for the prediction of SDHx mutations in paragangliomas/pheochromocytomas. We compared the mutation status of SDHx genes with the immunohistochemical (IHC) staining of SDH subunits in CPGLs. To identify pathogenic/likely pathogenic variants in SDHx genes, exome sequencing data analysis among 42 CPGL patients was performed. IHC staining of SDH subunits was carried out for all CPGLs studied. We encountered SDHx variants in 38% (16/42) of the cases in SDHx genes. IHC showed negative (5/15) or weak diffuse (10/15) SDHB staining in most tumors with variants in any of SDHx (94%, 15/16). In SDHA-mutated CPGL, SDHA expression was completely absent and weak diffuse SDHB staining was detected. Positive immunoreactivity for all SDH subunits was found in one case with a variant in SDHD. Notably, CPGL samples without variants in SDHx also demonstrated negative (2/11) or weak diffuse (9/11) SDHB staining (42%, 11/26). Obtained results indicate that SDH immunohistochemistry does not fully reflect the presence of mutations in the genes; diagnostic effectiveness of this method was 71%. However, given the high sensitivity of SDHB immunohistochemistry, it could be used for initial identifications of patients potentially carrying SDHx mutations for recommendation of genetic testing.

Currently, seven molecular subtypes of prostate cancer (PCa) are known, the most common of which being the subtype characterized by the presence of the TMPRSS2–ERG fusion transcript. While there is a considerable amount of work devoted to the influence of this transcript on the prognosis of the disease, data on its role in the progression and prognosis of PCa remain controversial. The present study is devoted to the analysis of the association between the TMPRSS2–ERG transcript and the biochemical recurrence of PCa. The study included two cohorts: the RNA–Seq sample of Russian patients with PCa (n = 72) and the TCGA–PRAD data (n = 203). The results of the analysis of the association between the TMPRSS2–ERG transcript and biochemical recurrence were contradictory. The differential expression analysis (biochemical recurrence cases versus biochemical recurrence-free) and the gene set enrichment analysis revealed a list of genes involved in major cellular pathways. The GNL3, QSOX2, SSPO, and SYS1 genes were selected as predictors of the potential prognostic model (AUC = 1.000 for a cohort of Russian patients with PCa and AUC = 0.779 for a TCGA–PRAD cohort).

Castration-resistant prostate cancer (CRPC) is a common form of prostate cancer in which docetaxel-based chemotherapy is used as the first line. The present study is devoted to the analysis of transcriptome profiles of tumor cells in the development of resistance to docetaxel as well as to the assessment of the combined effect with the XAV939 tankyrase inhibitor on maintaining the sensitivity of tumor cells to chemotherapy. RNA-Seq was performed for experimental PC3 cell lines as well as for plasma exosome samples from patients with CRPC. We have identified key biological processes and identified a signature based on the expression of 17 mRNA isoforms associated with the development of docetaxel resistance in PC3 cells. Transcripts were found in exosome samples, the increased expression of which was associated with the onset of progression of CRPC during therapy. The suppression of pathways associated with the participation of cellular microtubules has also been shown when cells are treated with docetaxel in the presence of XAV939. These results highlight the importance of further research into XAV939 as a therapeutic agent in the treatment of CRPC; moreover, we have proposed a number of mRNA isoforms with high predictive potential, which can be considered as promising markers of response to docetaxel.

Molecular heterogeneity in prostate cancer (PCa) is one of the key reasons underlying the differing likelihoods of recurrence after surgical treatment in individual patients of the same clinical category. In this study, we performed RNA-Seq profiling of 58 localized PCa and 43 locally advanced PCa tissue samples obtained as a result of radical prostatectomy on a cohort of Russian patients. Based on bioinformatics analysis, we examined features of the transcriptome profiles within the high-risk group, including within the most commonly represented molecular subtype, TMPRSS2-ERG. The most significantly affected biological processes in the samples were also identified, so that they may be further studied in the search for new potential therapeutic targets for the categories of PCa under consideration. The highest predictive potential was found with the EEF1A1P5, RPLP0P6, ZNF483, CIBAR1, HECTD2, OGN, and CLIC4 genes. We also reviewed the main transcriptome changes in the groups at intermediate risk of PCa—Gleason Score 7 (groups 2 and 3 according to the ISUP classification)—on the basis of which the LPL, MYC, and TWIST1 genes were identified as promising additional prognostic markers, the statistical significance of which was confirmed using qPCR validation.