Date of Award
Machine learning, Open reading frame, Pattern recognition, Prostate cancer biomarkers, Protein isoforms, RNA-Seq data
Prostate cancer is the most common cancer in men. One in eight Canadian men will be diagnosed with prostate cancer in their lifetime. The accurate detection of the disease’s subtypes is critical for providing adequate therapy; hence, it is critical for increasing both survival rates and quality of life. Next generation sequencing can be beneficial when studying cancer. This technology generates a large amount of data that can be used to extract information about biomarkers. This thesis proposes a model that discovers protein isoforms for different stages of prostate cancer progression. A tool has been developed that utilizes RNA-Seq data to infer open reading frames (ORFs) corresponding to transcripts. These ORFs are used as features for classificatio. A quantification measurement, Adaptive Fragments Per Kilobase of transcript per Million mapped reads (AFPKM), is proposed to compute the expression level for ORFs. The new measurement considers the actual length of the ORF and the length of the transcript. Using these ORFs and the new expression measure, several classifiers were built using different machine learning techniques. That enabled the identification of some protein isoforms related to prostate cancer progression. The biomarkers have had a great impact on the discrimination of prostate cancer stages and are worth further investigation.
Alshehri, Manal, "A Machine Learning Model for Discovery of Protein Isoforms as Biomarkers" (2016). Electronic Theses and Dissertations. 5900.