A network-based, integrative study to identify core biological pathways that drive breast cancer clinical subtypes.

Background: The rapid collection of diverse genome-scale data raises the urgent need to integrate and utilise these resources for biological discovery or biomedical applications. For example, diverse transcriptomic and gene copy number variation data are currently collected for various cancers, but relatively few current methods are capable to utilise the emerging information.

Methods: We developed and tested a data-integration method to identify gene networks that drive the biology of breast cancer clinical subtypes. The method simultaneously overlays gene expression and gene copy number data on protein-protein interaction, transcriptional-regulatory and signalling networks by identifying coincident genomic and transcriptional disturbances in local network neighborhoods.

Results: We identified distinct driver-networks for each of the three common clinical breast cancer subtypes: oestrogen receptor (ER)+, human epidermal growth factor receptor 2 (HER2)+, and triple receptor-negative breast cancers (TNBC) from patient and cell line data sets. Driver-networks inferred from independent datasets were significantly reproducible. We also confirmed the functional relevance of a subset of randomly selected driver-network members for TNBC in gene knockdown experiments in vitro. We found that TNBC driver-network members genes have increased functional specificity to TNBC cell lines and higher functional sensitivity compared with genes selected by differential expression alone.

Conclusion: Clinical subtype-specific driver-networks identified through data integration are reproducible and functionally important.