Table 3. Linear models of transcriptional noise with genomic and epigenomic factors.

	OLS			GLS
	Coefficient	SE	P-value	Coefficient	SE	P-value
(Intercept)	0.1612	0.0781	0.0392*	0.1665	0.0663	0.0121*
PC1	0.0390	0.0065	< 0.0001***	0.0396	0.0065	< 0.0001***
PC2	−0.0048	0.0069	0.4854	−0.0048	0.0069	0.4838
PC3	−0.0526	0.0091	< 0.0001***	−0.0518	0.0092	< 0.0001***
PC4	−0.0102	0.0097	0.2905	−0.0109	0.0100	0.2773
PC5	0.0117	0.0106	0.2713	0.0123	0.0106	0.2456
PC6	−0.0152	0.0107	0.1536	−0.0152	0.0109	0.1623
PC7	0.0210	0.0102	0.0384*	0.0211	0.0110	0.0561.
PC8	0.0100	0.0113	0.3778	0.0073	0.0114	0.5250
TFPC1	0.0028	0.0041	0.4912	0.0025	0.0034	0.4658
TFPC2	0.0025	0.0027	0.3664	0.0024	0.0026	0.3585
TFPC3	0.0032	0.0042	0.4513	0.0032	0.0037	0.3825
HistPC1	−0.0031	0.001	0.0015**	−0.0033	0.0010	0.0007***
HistPC2	−0.0027	0.0016	0.0846.	−0.0029	0.0015	0.0566.

All correlations are computed using Kendall’s rank correlation test, with P-value codes defined as *** < 0.001 < ** < 0.01 < * < 0.05 < ^. < 0.1. OLS, Ordinary Least Squares; GLS, Generalized Least Squares; Pathway PC1–8, principal components on centrality measures, protein conservation, and gene age; TFPC1–3, principal components of the logistic PCA on transcription factor binding evidence; HistPC1 and 2, principal components of the logistic PCA on histone modification marks.