Oncoprotein-specific molecular interaction maps (SigMaps) for cancer network analyses.

Area of Science:

• Oncology
• Systems Biology
• Bioinformatics

Background:

• Current cancer pathway models lack specificity for individual oncoproteins and cellular contexts.
• Understanding oncoprotein interactions is crucial for cancer mechanism elucidation and predicting therapeutic responses.

Purpose of the Study:

• To develop context-specific oncoprotein interaction networks (SigMaps) for improved cancer research.
• To create a generalizable machine learning framework (OncoSig) for reconstructing these networks.

Main Methods:

• Reconstruction of SigMaps by integrating diverse data sources including protein structure, gene expression, and mutational profiles.
• Utilizing the OncoSig machine learning framework for de novo network generation.
• Experimental validation of identified synthetic lethal proteins in 3D spheroid models.

Main Results:

• A KRAS-specific SigMap for lung adenocarcinoma was generated, confirming known KRAS biology and identifying novel synthetic lethal partners.
• Demonstrated crosstalk between KRAS and RAB/RHO signaling pathways.
• Generated SigMaps for the ten most mutated human oncoproteins and all 715 proteins in the COSMIC Cancer Gene Census.

Conclusions:

• SigMaps provide context-specific insights into oncoprotein interactions and cellular regulatory architecture.
• The OncoSig framework is generalizable across various oncoproteins and cancer types.
• Cancer's signaling architecture exhibits significant tissue specificity, highlighting the importance of context-specific analysis.