[back to the DMCM’25 Programme]

Title:
lung.cRegMap: a web platform for unraveling the lung cancer coregulatory network and identifying potential key regulators

Authors:
Geoffrey Pawlak, David Tulasne, and Mohamed Elati

Abstract:
Lung cancer research, driven by numerous large-scale studies conducted across various laboratories, has generated an extensive volume of data from both in vitro and in vivo experiments. Consolidating these datasets into a unified and accessible system-level format is pivotal for encompassing a broader spectrum of tumor types and represents a critical step towards achieving a comprehensive understanding of lung cancer.

Using transcriptomic data from lung cancer cell lines provided by the depmap R package, we inferred a highly specific lung cancer gene regulatory network (GRN) employing the hLICORN algorithm and the CoRegNet R package. We developed lung.cRegMap, an open-source and user-friendly web tool, in which we encapsulated our lung-specific GRN. Lung.cRegMap features three components: “CoRegNet,” visualizing the GBM-CoRegNet network via visNetwork for intuitive exploration of phenotype-specific coregulator colocalization. “CoRegMap,” a unified influence map of consolidated tumors annotated with molecular, clinical, and mutational data to align and colocalize patient-derived cell models. Finally, “CoRegQuery,” enabling the mapping of user-uploaded, through a CSV file or the Gene Expression Omnibus (GEO)-derived expression datasets by supplying a GEO accession code from which GBM-cRegMap extracts the gene expression dataset and experimental information. Data in CSV format or GEO IDs are processed to compute regulator influences, facilitating the investigation of coregulators, interactions, and systems. A machine learning module enhances the prediction of query sample localization within GBM-CoRegMap, ensuring consistent comparisons of tumor transcriptional patterns and enabling deeper exploration of tumor heterogeneity and phenotype adaptability. Generated results are easily downloadable for further research purposes. We showcase the capability of cRegMap to comprehensively analyze lung cancer heterogeneity using a meta-cohort of 42 datasets encompassing 5,487 patients eliminating batch-effect. Through the synergistic integration of its components, cRegMap refines existing lung cancer classifications, defines novel subgroups, characterize them through clinical, mutational and gene dependency projection, identifies potential key regulators, and facilitates the alignment of tumors with in vitro models.

Lung.cRegMap web application aims to unravel lung cancer heterogeneity and plasticity, offering a unique framework to uncover latent molecular mechanisms and advancing cancer biology and precision oncology.