Jmp 13.2.1 statistical software

Patients were dichotomized into HR and LR disease if they had a distant metastasis or death before 5 years. Differentially expressed genes (DEGs) were identified by performing multiple T-test corrected by the Benjamini–Hochberg procedure with an FDR of 0.1, using JMP 13.2.1 statistical software (SAS) and correlation between survival time in months and gene expression was also determined using JMP. Matlab statistics and machine learning toolbox was used to identify features (i.e., genes) using the univariate feature ranking function based on chi-square tests, or features were manually selected based on their ranks following differential analysis of gene expression (DEG) or correlation as described above, with a fourth list comprised of genes in common following DEG and correlation. For training, 89/99 samples were randomly chosen so as to leave out 10 samples (5 aggressive and 5 non-aggressive) for subsequent validation. Three models were chosen from the Matlab menu for training and validation, which included Linear Discriminant Analysis, Logistic Regression (LR), and Support Vector Machine (quadratic SVM). For validation, sensitivity was defined as number of true aggressive tumors predicted divided by total number of true aggressive tumors, and specificity defined as true negatives predicted divided by all true negatives.