Table 2.

Associations between COGs and descriptive variables.

COG	Number of gene analyzed	Bonferroni-corrected P value for one-sided U-test testing for associations between genes in a given COG and(1)
		> length	> nt diversity	> Number of Informative sites	> Codon bias(2)	< Codon bias(2)
Energy production and conversion	100	< 0.002	ND	NS	0.007	ND
Cell cycle control, mitosis and meiosis	20	NS	ND	ND	NS	ND
Amino acid transport and metabolism	181	< 0.002	NS	< 0.002	NS	ND
Nucleotide transport and metabolism	63	NS	ND	NS	NS	ND
Carbohydrate transport and metabolism	186	< 0.002	ND	NS	NS	ND
Coenzyme transport and metabolism	87	NS	< 0.001	0.004	ND	0.047
Lipid transport and metabolism	51	NS	NS	NS	NS	ND
Translation	91	NS	ND	ND	< 0.001	ND
Transcription	187	ND	ND	ND	ND	< 0.002
Replication, recombination and repair	92	< 0.002	< 0.001	< 0.002	ND	NS
Cell wall/membrane biogenesis	76	< 0.002	< 0.001	< 0.002	NS	ND
Cell motility	40	NS	ND	NS	ND	NS
Posttranslational modification, protein turnover, chaperones	52	ND	ND	ND	0.021	ND
Inorganic ion transport and metabolism	104	NS	ND	NS	ND	NS
Secondary metabolites biosynthesis, transport and catabolism	31	ND	NS	NS	ND	NS
General function prediction only	254	NS	NS	0.008	ND	NS
Function unknown	161	ND	NS	ND	ND	NS
Signal transduction mechanisms	104	NS	ND	NS	ND	< 0.002
Intracellular trafficking and secretion	37	NS	ND	ND	NS	ND
Defense mechanisms	51	< 0.002	NS	0.020	ND	NS
Not in COGs	511	ND	ND	ND	ND	NS

⁽¹⁾ ">" or "<" indicates the direction of the one-sided tests (i.e. column "> Codon bias" shows Bonferroni-corrected p-values for associations between genes in a given COG and higher codon bias (as compared to the genes in other COGs), while the column "< Codon bias" test shows Bonferroni-corrected p-values for genes in COGs with lower codon bias as compared to genes in other COGs; "ND" = Not determined (tests were not performed for COGs that showed values that were not consistent with the tested alternative hypothesis, e.g., if the average gene length for genes in a given COG was below average than we did not test for an association of this COG with increased gene length); "NS". Not significant.

⁽²⁾Tests for codon bias were performed using Nc values; a lower Nc indicates increased codon bias.