Table 2.

Parameters used in each method.

Feature selection algorithms
Algorithm	Parameters
ttest	Default parameters
Mann_Whitney	Default parameters
DCor	Default parameters
Boruta	{perc: 100, max_iter: 100, n_estimators: 15,000, max_depth: 6}
Lasso	{alpha: 0.001, max_iter: 20,000}
Lasso	{alpha: 0.01, max_iter: 20,000}
ElasticNet	{l1_ratio: 0.5, max_iter: 20,000, alpha: 0.001}
ElasticNet	{l1_ratio: 0.5, max_iter: 20,000, alpha: 0.01}
RandomForestClassifier	{n_estimators: 10,000, max_depth: null}
RidgeCV	default parameters
SVM(SVC)	{kernel: linear, C: 1}
Recursive feature selection with random forest	{n_estimators: 500, max_depth: null}
Recursive feature selection with SVM (SVC)	{kernel = linear}

Class prediction algorithms
Algorithm	Parameters
RandomForestClassifier	{n_estimators = 1000, max_depth = 4}
SVC	{C = 1, kernel = 'linear'}
LogisticRegression	{max_iter = 20,000}
Lasso	{max_iter = 20,000, alpha = .001}
ElasticNet	{max_iter = 20,000, alpha = .001, l1_ratio = .5}