Methodology for biomarker discovery with reproducibility in microbiome data using machine learning

BMC Bioinformatics

Table 1 Individual and average AUCs and MCCs from the validation phase and the additional validation approaches applied to the ASD datasets. The standard deviation of each result was excluded to keep the table simple and avoid complexity

David et al.*	26 features (REFS)		2040 features		SelectKbest (k = 26)
Classifier	AUC	MCC	AUC	MCC	AUC	MCC
AdaBoostClassifier	0.7200	0.4355	0.3900	− 0.1726	0.7500	0.5133
Extra Trees	0.7800	0.5934	0.3400	− 0.3664	0.7400	0.5195
KNeighbors	0.7900	0.6407	0.4200	0.0486	0.6200	0.2468
MLP	0.9000	0.8549	0.4100	− 0.0709	0.7500	0.3934
Lasso CV	0.8900	0.7207	0.5000	− 0.0447	0.6700	0.3177
Average	0.8160	0.6490	0.4100	− 0.1212	0.7060	0.3981

PRJNA589343	22 of 26 features (REFS)		SelectKbest (20 of 26 features)		10-time random selection
Classifier	AUC	MCC	AUC	MCC	AUC	MCC
AdaBoostClassifier	0.7700	0.5968	0.6460	0.5139	0.7700	0.2474
Extra Trees	0.8400	0.7208	0.6510	0.6919	0.8000	0.3205
KNeighbors	0.6800	0.4808	0.6220	0.3967	0.6800	0.2636
MLP	0.7400	0.5119	0.6490	0.4181	0.7300	0.2863
Lasso CV	0.7100	0.0867	0.5710	0.1877	0.5400	0.1320
Average	0.7480	0.4794	0.6278	0.4416	0.7040	0.2500

PRJNA578223	20 of 26 features (REFS)		SelectKbest (17 of 26 features)		10-time random selection
Classifier	AUC	MCC	AUC	MCC	AUC	MCC
AdaBoostClassifier	0.8300	0.7089	0.6530	0.2725	0.6700	0.3359
Extra Trees	0.8400	0.7105	0.6510	0.3877	0.6900	0.3578
KNeighbors	0.7000	0.2570	0.6370	0.2924	0.5900	0.3318
MLP	0.7200	0.4816	0.6120	0.4576	0.7300	0.2398
Lasso CV	0.6100	0.3779	0.6230	0.5025	0.7100	0.2738
Average	0.7400	0.5071	0.6352	0.3825	0.6780	0.3078

ISSN: 1471-2105