Machine learning-based predictions of dietary restriction associations across ageing-related genes

BMC Bioinformatics

Table 2 Sensitivity and specificity values across all {ML algorithm, Dataset} models

Feature type	Num. of features	Num. of instances	Sensitivity				Specificity				Mean
Feature type	Num. of features	Num. of instances	BRF	EEC	XGB	CAT	BRF	EEC	XGB	CAT	Sens	Spec
PathDIP	1640	986 genes	0.87	0.77	0.58	0.77	0.67	0.7	0.93	0.77	0.75	0.77
KEGG-Pertinence	312	799 genes	0.74	0.70	0.75	0.72	0.7	0.71	0.76	0.73	0.73	0.73
KEGG-Influence	1770	799 genes	0.75	0.70	0.67	0.67	0.66	0.71	0.69	0.68	0.70	0.69
PPI-measures	18	850 genes	0.64	0.62	0.6	0.57	0.65	0.57	0.71	0.77	0.61	0.68
PPI-adjacency	5718	850 genes	0.65	0.52	0.5	0.65	0.62	0.74	0.78	0.67	0.58	0.70
GO terms	8640	1124 genes	0.85	0.80	0.48	0.69	0.67	0.72	0.94	0.78	0.71	0.78
GTEx	55	1111 genes	0.54	0.55	0.45	0.57	0.48	0.51	0.57	0.48	0.53	0.51
Co-expression	44,946	1048 genes	0.58	0.61	0.21	0.56	0.45	0.86	0.85	0.54	0.49	0.68
Proteins-Descriptors	156	6180 Proteins (from 1109 genes)	0.25	0.43	0.45	0.48	0.86	0.91	0.88	0.77	0.40	0.86
Whole-Dataset Imputation	63,099	1137 genes	0.69	0.68	0.43	0.71	0.65	0.63	0.87	0.64	0.63	0.70
Whole-Dataset Intersection	63,099	628 genes	0.60	0.72	0.39	0.75	0.63	0.62	0.87	0.56	0.62	0.67
Mean	–	–	0.65	0.65	0.50	0.65	0.64	0.70	0.80	0.67	0.61	0.70

Sensitivity (\(Ageing_{DR}\) prediction quality) and specificity (\(Ageing_{NotDR}\) prediction quality) values for the BRF, EEC, XGB, and CAT algorithms across all the datasets (feature types). The average sensitivity and specificity of each algorithm across all the datasets is shown in the last row; and the average Gmean and AUC of each feature type across all four algorithms is shown in the last two columns

ISSN: 1471-2105