AI-Driven Identification and Validation of RXFP1 as a CML Biomarker Using Gene Expression and Integrated GUI Tools

Abstract

Chronic myeloid leukemia (CML) is a hematologic malignancy primarily driven by the BCR-ABL1 fusion gene. While tyrosine kinase inhibitors (TKIs), such as imatinib, have transformed CML management, many patients still experience adverse side effects and therapy resistance. This study investigated the potential of RXFP1 (relaxin family peptide receptor 1) as a diagnostic biomarker for CML, leveraging bioinformatics, statistical analysis, and artificial intelligence (AI)-based modeling. Using the GSE97562 gene expression dataset, which consists of 40 human bone marrow samples, we identified RXFP1 as the top-ranked gene in a random forest model trained to differentiate CML samples from normal samples. The gene exhibited statistically significant overexpression in CML (p < 0.0001), with an area under the ROC curve (AUC) of 1.0, indicating perfect classification capability. Notably, RXFP1 retained its expression profile even after imatinib treatment, suggesting its independence from BCR-ABL signaling and making it a promising candidate for disease monitoring regardless of therapeutic status. To facilitate nonprogrammatic analysis and reproducibility, a GUI tool was developed using Python, which integrates statistical tests, visual plots (ROC, box plots, Mann-Whitney curve), and a text-to-speech reporting system. Manual validation further confirmed RXFP1s discriminatory power, with zero expression overlapping between CML and normal samples across all experimental subgroups. This research is also driven by a personal motivation: the authors wife is a CML patient undergoing TKI treatment and experiencing side effects. This study emerged from the authors desire to find a safer and more precise diagnostic pathway. These findings suggest that RXFP1 is a robust and treatment-independent molecular biomarker for CML, laying a foundation for larger studies and therapeutic exploration. This study also demonstrates the potential of integrated AI tools in accelerating biomarker discovery and empowering independent researchers to contribute to precision medicine.