Fig. 5

Pipeline to test the performance of the shape-based assumption in patient survival prediction. This pipeline is used once for each assumption: shape, symmetric and random where the difference occurs in step 2. For the shape assumption, we will separate the samples using the shape (Fig. 4a–c) and then compute the p value with a log-rank test. For the symmetric assumption, we will separate the samples into two groups using the same splitting rule for all the genes (Fig. 4a) and then compute the p value with a log-rank test. For the random assumption, the 100 genes are just chosen at random from the dataset and then compute the p value with a log-rank test