Genelens: A Python Package Implementing Monte Carlo Machine Learning and Network Analysis Methods for Biomarker Discovery and Gene Functional Annotation

We present GeneLens, a Python package for comprehensive analysis of differentially expressed genes and biomarker discovery. The package consists of two core modules: FSelector for biomarker identification by utilizing Monte Carlo simulations of L1-regularized models, and NetAnalyzer for functional prediction of selected gene sets based on the topology of their protein-protein interaction networks.The FSelector includes: (1) automated gene selection through iterative bootstrap sampling; (2) calculation of gene significance weights taking into account ROC-AUC model performance and their number in simulations; (3) adaptive thresholding for feature space reduction. NetAnalyzer performs pathway enrichment analysis while integrating significance weights from FSelector. Implemented as a PIP module, GeneLens provides standardized algorithms for applying machine learning and network analysis methods in differential gene expression studies, along with automated model hyperparameter tuning and visualization tools.

transcriptomics, machine learning, Monte Carlo, biomarkers, differentially expressed genes, network analysis