Microrna - based drug repurposing analysis in glioblastoma multiforme via machine learning approaches

Glioblastoma (GBM) is the most common cancer among all brain tumours and has the worst prognosis. In this thesis, we applied various bioinformatic analyses and machine learning classification techniques to GBM data to illustrate and better understand the biomolecular mechanisms of GBM. First, GBM datasets to be analysed were selected. Then, expression analyses of transcriptome and miRNA data were performed. Differentially expressed genes and miRNAs were identified, gene clusters co-expressed with common genes and genes targeted by miRNAs were identified, and the co-expression network was drawn. The prognostic properties of these gene clusters were examined and performance analyses were performed with machine learning classification algorithms such as Random Forest (RF), Support Vector Machine (SVM), Naive Bayes (NB) and The K-Nearest Neighbours (KNN). As a result of the performance analyses, RF, SVM and KNN algorithms were found to be reliable and accurate classification algorithms that work with existing gene groups with accuracy scores above 80%. In addition, according to the results obtained by upward and downward feature selection, it was determined that SEPT4, VAMP1, MAP1A, KIF5C, NPTX1 and ATP8A1 genes give high interaction and may be important biomarker candidates in GBM disease. In addition, the gene clusters obtained were used for drug repositioning analyses. As a result of these tests, the gene groups with the highest interactions in the gene clusters belonging to the gene co-expression network module and miRNA-based co-expression network module data in machine learning analyses and their common drugs were determined. A large number of drug candidates such as Emetine Dihydrochloride Hydrate (74), 16beta Bromoandrosterone, AS605240, 480743.cdx, BRD K00627859, HDAC6 inhibitor ISOX, BRD-K12184916 and 16-HYDROXYTRIPTOLIDE were found for the treatment of GBM. It was determined that these drugs and small molecules can be tested in future clinical trials and may be new drug candidates for the treatment of GBM.