Table 3.
MSigDB C2 Gene Set (Source) | (A) Gene set enrichment | (B) GSEA in meta-analyzed case vs control gene expression data | |||||
---|---|---|---|---|---|---|---|
# cis-genes | p-value (random) | p-value (structured) | Skeletal Muscle | Adipose | Liver | Pancreas | |
BCAA degradation | 5 | 0.020 | 0.027 | ↓<2.5E-04 | ↓1.7E-04 | ↓6.0E-03 | ↓2.5E-04 |
Biotin carboxylases | 2 | 0.022 | 0.019 | ↓9.0E-03 | ↓2.5E-04 | 0.20 (n.s.) | ↓7.5E-04 |
Propanoate metabolism | 4 | 0.025 | 0.042 | ↓<2.5E-04 | ↓1.7E-04 | 0.13 (n.s.) | ↓3.3E-04 |
Butanoate metabolism | 4 | 0.032 | 0.042 | ↓9.0E-03 | ↓1.7E-04 | 0.14 (n.s.) | ↓2.5E-04 |
(A) Four curated mitochondrial gene sets showed evidence of enrichment by count among the total 763 T2D cis-genes (p-value<0.05), compared to the genomic background. All gene sets were curated from the KEGG database excluding biotin carboxylases which was manually defined. Empirical p-values were calculated using a permutation approach by selecting genes at random from the genomic background (random) or from within ±1.5Mb of each other (structured) (the distance used to investigate putative cis-genes). (B) FDR values from GSEA for the same four gene sets, showing significant enrichment for decreased expression in meta-analyzed case-control gene expression datasets. ↓ indicates a significant enrichment for decreased expression compared to the background and no arrow indicates an enrichment for mixed differential expression (both increased and decreased). n.s., not significant (p-value > 0.05).