Table 5. Summary of results from the GO enrichment and feature importance analyses on the genes underexpressed with age.

	All tissues (Underexpressed)
	GO.ID	Term	p-value	Precision (0.0090)
	GO:0010510	Regulation of acetyl-CoA biosynthetic process from pyruvate	3.20e-6	0.374
	GO:0006122	Mitochondrial electron transport, ubiquinol to cytochrome c	1.60e-9	0.356
	GO:0006099	Tricarboxylic acid cycle	1.50e-10	0.281
	GO:0006107	Oxaloacetate metabolic process	1.20e-4	0.217
	GO:0007528	Neuromuscular junction development	3.06e-2	0.204
	Brain (Underexpressed)
	GO.ID	Term	p-value	Precision (0.0008)
	GO:0021782	Glial cell development	3.69e-2	0.0489
	GO:0021510	Spinal cord development	4.20e-2	0.0425
	Muscle (Underexpressed)
	GO.ID	Term	p-value	Precision (0.0041)
	GO:0043455	Regulation of secondary metabolic process	1.16e-3	0.203
	GO:0006094	Gluconeogenesis	1.50e-9	0.170
	GO:0061621	Canonical glycolysis	4.30e-6	0.146
	GO:0042776	Mitochondrial ATP synthesis coupled proton transport	4.90e-5	0.119
	GO:0006107	Oxaloacetate metabolic process	9.70e-4	0.103

Presented here are a selection of terms for each tissue which were both significantly enriched in the given gene list and present in the top-20 terms, ranked by precision, for the prediction of a gene as being underexpressed by the Random Forest model. The value given between brackets in the Precision column header is the class label’s relative class frequency, i.e. the precision that a classifier would get by randomly classifying the genes. It should be noted that the list of genes underexpressed in the heart was too small for a meaningful analysis and so has been left out.