Grant: Doctoral Research Award: Canada Graduate Scholarships

The modifications in exonic regions of DNA (parts of DNA sequence that are translated to proteins) can affect the structure and consequently the function of proteins and the cellular pathways they participate in. Changes in the cellular network can be detected in the transcriptome, i.e., the set of all RNA molecules expressed by a cell or a set of cells. It is vital to detect specific cellular modifications in individuals with cancer since these changes affect the response to different therapies. Additionally, the adverse effects of drugs, including resistance to therapy, emerge only in a subset of patients. These factors necessitate individualized therapies to improve patient outcomes. Machine learning (ML) algorithms, which are powerful tools that automate analytical model building, have been successfully used to classify cancer types, find informative features in cancer diagnosis, predict drug sensitivity markers, and accomplish many other goals. A benefit of utilizing ML algorithms has also been illustrated in categorizing tumour samples based on cellular pathways activities. This project aims to find active and targetable cellular pathways through investigating tumour samples transcriptome using ML approaches. We intend to classify the tumour samples based on the presence or absence of a mutation in the genes that play a crucial role in cancer initiation and progression, and analyze the transcriptional patterns identified by the algorithm. Our initial results demonstrated that ML algorithms such as random forests could successfully classify tumour samples based on the occurrence of tumour protein (TP53) gene mutations and output genes and proteins essential in classification. These findings will be leading to identifying targetable pathways and distinguishing patients who can benefit from targeted therapies.