Abstract:
Introduction: Discovering the players and understanding the molecular mechanisms involved in cancer initiation and progression are fundamental issues for developing effective diagnostic tools and targeted therapies. Traditional methods for predicting protein involvement in cancer usually depend on identifying aberrant mutations in individual genes or linking transcript levels to patient survival. These methods are usually carried out separately and concentrate on one gene/protein at a time, ignoring nucleotide changes occurring outside of coding regions or mutations happening simultaneously in genes within the same network of interactions.
Methods: Here, by crossing publicly available mutational datasets with clinical outcome prediction and interactomic data, we constructed an R pipeline capable to define novel protein hubs predicted to play important roles in cancer. By adopting this comprehensive strategy, which we defined CancerHubs, we were able to rank genes according to a newly introduced metric termed the 'network score'. The network score predicts the level of involvement of a certain gene in a particular cancer by defining the number of mutated interactors that its encoded protein has.
Results: Taking advantage of the CancerHubs approach, we identified several novel broad-cancer and cancer-specific genes. Among these, we validated two: one encoding for a protein with broad tumour suppressor functions, and one encoding for a protein with oncogenic properties in Multiple Myeloma. Our findings underline the importance of considering diverse molecular data types and network-level interactions in order to fully unravel the complexity of cancer biology and pinpoint novel potential therapeutic targets.
Conclusions: CancerHubs introduces a pioneering method for forecasting gene involvement in cancer. By ranking cancer-associated genes based on the number of mutant interactors their encoded proteins have, this approach identifies protein hubs potentially involved in cancer. This methodology globally improves the overall detection of cancer-associated genes, as demonstrated by its ability to accurately predict protein hubs previously never found related to cancer.
Audience Take Away Notes:
- CancerHubs is a method that, by combining in a totally novel way unbiased mutational data, clinical outcome predictions and interactomic data, predicts if, and to what extent, cancer-related proteins are part of more broad cancer-mutated networks.
- By using the CancerHubs approach scientists can generate lists of genes with putative impact in cancer pathology and from such lists they can define novel broad-cancer or cancer-specific genes.
- Definition of novel cancer-related genes/pathways can deeply influence cancer research, paving the way for novel and more efficient therapeutic approaches.
