"The Logistic Regression model was chosen from the set of tested models and Matthew's Correlation Coefficient (MCC) was chosen as most appropriate scorer. This block performs an automatic feature selection and ranking, followed by an optimization of the hyperparameters $L$ (norm for penalization) and $C$ (inverse regularization strength) with respect to MCC. The optimized Logistic Regression Model is then assessed as described in Step 4 above. Note that the model optimization is done using a nested cross-validation scheme as explained in section 3.3 (Figure 4a).<br>\n",

"The Logistic Regression model was chosen from the set of tested models and Matthew's Correlation Coefficient (MCC) was chosen as most appropriate scorer. This block performs an automated feature selection and ranking, followed by an optimization of the hyperparameters $L$ (norm for penalization) and $C$ (inverse regularization strength) with respect to MCC. The optimized Logistic Regression Model is then assessed as described in Step 4 above. Note that the model optimization is done using a nested cross-validation scheme as explained in section 3.3 (Figure 4a).<br>\n",

"One table and seven figures are produced in this block:\n",

"+ Table c: Cross-validated performance scores of final model (Table B1).\n",

"+ Figure e: Result of automatic feature selection (Figure 4b). The numbers at the top indicate the IDs of the features (consistent with their position in array \"Features\") ordered from the most important one (left) to the least important one (right, numers at the bottom).\n",

"+ Figure e: Result of automated feature selection (Figure 4b). The numbers at the top indicate the IDs of the features (consistent with their position in array \"Features\") ordered from the most important one (left) to the least important one (right, numers at the bottom).\n",

"+ Figure f: Result of model optimization (Figure S2).\n",