Table 1.

Carbon content biases of proteins encoded by single-copy (S) and duplicated (D) genes in different protein functional categories. (Column 2 shows the number of analysed proteins, n, in each category, separately, for S and D genes. Column 3 shows the Spearman's rank correlation coefficients, r_s (CAI), between carbon content per amino acid and CAI. Column 4 shows the difference in median carbon content between S and D genes. p-values in parentheses are derived from Mann–Whitney U-tests. Values in bold indicate p<0.01 for the difference between carbon cost of S and D genes. Note that duplicate genes always have lower C content, regardless of category. Column 5 shows p-values (P(r)) for Mann–Whitney U-tests of residuals from linear regressions between carbon content and (log) CAI, for all categories where the association between CAI and carbon content was significant. Differences here indicate that high expression (CAI) cannot fully explain carbon cost differences between S and D genes (^*p<0.05, ^**p<0.01, ^***p<0.001).)

function	n (S, D)	rs (CAI)	median S−D (×10−2)	p	P(r)
metabolism or energy	1050, 595a	−0.147***	3.10	(<10−3)	0.010
cell cycle and DNA processing	793, 200a	−0.064*	4.96	(<10−3)	0.001
transcription	844, 178	0.025	2.92	(0.023)
translation	330, 147	−0.257***	4.90	(0.006)	0.516
protein fate	864, 281a	−0.077**	4.31	(0.001)	0.007
cellular transport, transport facilitation, transport routes	703, 325	−0.089**	2.40	(0.006)	0.046
cell rescue defence and virulence	337, 215	−0.227***	1.70	(0.112)	0.781
biogenesis of cellular components	659, 197	−0.092**	6.21	(<10−3)	0.041
cell type differentiation	301, 146	0.004	2.31	(0.180)

^aOne duplicated CDS did not have a CAI value available (analyses of CAI have n−1 data points).