MBMethPred: a computational framework for the accurate classification of childhood medulloblastoma subgroups using data integration and AI-based approaches.

Childhood medulloblastoma is a malignant form of brain tumor that is widely classified into four subgroups based on molecular and genetic characteristics. Accurate classification of these subgroups is crucial for appropriate treatment, monitoring plans, and targeted therapies. However, misclassification between groups 3 and 4 is common. To address this issue, an AI-based R package called MBMethPred was developed based on DNA methylation and gene expression profiles of 763 medulloblastoma samples to classify subgroups using machine learning and neural network models. The developed prediction models achieved a classification accuracy of over 96% for subgroup classification by using 399 CpGs as prediction biomarkers. We also assessed the prognostic relevance of prediction biomarkers using survival analysis. Furthermore, we identified subgroup-specific drivers of medulloblastoma using functional enrichment analysis, Shapley values, and gene network analysis. In particular, the genes involved in the nervous system development process have the potential to separate medulloblastoma subgroups with 99% accuracy. Notably, our analysis identified 16 genes that were specifically significant for subgroup classification, including EP300, CXCR4, WNT4, ZIC4, MEIS1, SLC8A1, NFASC, ASCL2, KIF5C, SYNGAP1, SEMA4F, ROR1, DPYSL4, ARTN, RTN4RL1, and TLX2. Our findings contribute to enhanced survival outcomes for patients with medulloblastoma. Continued research and validation efforts are needed to further refine and expand the utility of our approach in other cancer types, advancing personalized medicine in pediatric oncology.

The cervical transcriptome changes during the menstrual cycle but does not predict the window of implantation.

Introduction: The expression of genes in female reproductive organs is influenced by the cyclic changes in hormone levels during the menstrual cycle. While the molecular changes in the endometrium that facilitate embryo implantation have been extensively studied, there is limited knowledge about the impact of the menstrual cycle on cervical cells. Cervical cells can be easily and routinely collected using a cytobrush during gynecological examination, offering a standardized approach for diagnostic testing. In this study we investigated how the transcriptome of cervical cells changes during the menstrual cycle and assessed the utility of these cells to determine endometrial receptivity. Methods: Endocervical cells were collected with cytobrushes from 16 healthy women at different menstrual cycle phases in natural cycles and from four women undergoing hormonal replacement cycles. RNA sequencing was applied to gain insight into the transcriptome of cervical cells. Results: Transcriptome analysis identified four differentially expressed genes (DEGs) between early- and mid-secretory samples, suggesting that the transcriptome of cervical cells does not change significantly during the opening of the implantation window. The most differences appeared during the transition to the late secretory phase (2136 DEGs) before the onset of menstruation. Cervical cells collected during hormonal replacement cycles showed 1899 DEGs enriched in immune system processes. Conclusions: The results of our study suggested that cervical cells undergo moderate transcriptomic changes throughout the menstrual cycle; however, these changes do not reflect the gene expression pattern of endometrial tissue and offer little or no potential for endometrial receptivity diagnostics.

The expression pattern of endometrial receptivity genes is desynchronized between endometrium and matched endometriomas.

RESEARCH QUESTION: Are paired samples of endometrium and ovarian endometriomas synchronous with each other throughout the menstrual cycle? DESIGN: The expression levels of 57 endometrial receptivity-associated genes were determined from matched endometrial and endometrioma samples (n=31) collected from women with endometriosis throughout the menstrual cycle. RESULTS: The expression profile of endometrial receptivity genes divided endometrial samples according to their menstrual cycle phase. Endometrioma samples grouped together irrespective of the menstrual cycle phase and formed a cluster distinct from endometrial samples. Pairwise comparison showed 21, 16, 33 and 23 differentially expressed genes (adjusted P < 0.001-0.05) between the lesions and endometria collected in the proliferative, early-secretory, mid-secretory and late-secretory menstrual cycle phases, respectively, confirming the distinct expression profiles of endometrium and endometrioma. CONCLUSIONS: No menstrual cycle synchronicity was found between matched eutopic and ectopic endometrium, suggesting that the concept of cycling endometrial tissue inside the endometrioma should be revised.

Machine Learning Approaches to Classify Primary and Metastatic Cancers Using Tissue of Origin-Based DNA Methylation Profiles.

Metastatic cancers account for up to 90% of cancer-related deaths. The clear differentiation of metastatic cancers from primary cancers is crucial for cancer type identification and developing targeted treatment for each cancer type. DNA methylation patterns are suggested to be an intriguing target for cancer prediction and are also considered to be an important mediator for the transition to metastatic cancer. In the present study, we used 24 cancer types and 9303 methylome samples downloaded from publicly available data repositories, including The Cancer Genome Atlas (TCGA) and the Gene Expression Omnibus (GEO). We constructed machine learning classifiers to discriminate metastatic, primary, and non-cancerous methylome samples. We applied support vector machines (SVM), Naive Bayes (NB), extreme gradient boosting (XGBoost), and random forest (RF) machine learning models to classify the cancer types based on their tissue of origin. RF outperformed the other classifiers, with an average accuracy of 99%. Moreover, we applied local interpretable model-agnostic explanations (LIME) to explain important methylation biomarkers to classify cancer types.