Fig. 3

The gene ontology analysis of transcripts ranked by their codon composition (proportion of GC3 or AU3 codons) identifies several categories that can be divided in two subgroups of genes. a. Several significant GO categories can be identified on the basis of the percentages of the GC3 codons of human mRNAs. For this analysis we used the WebGestalt 2017 gene set enrichment analysis toolkit [22]. See Methods for details. b. Although the gene set enrichment analysis of WebGestalt discards results obtained by chance based on the FDR, we wanted to make sure that the categories identified here truly depend on their GC3 codon percentages. For this purpose, we repeated the analysis after randomly changing the percentage of the GC3 codons for each coding sequence, and we counted the number of gene ontology categories that were identified with a false discovery rate (FDR) below 0.01, in at least 5 different replicates. Changing the average GC3 percentage by as little as 5% already reduced the number of GO classes that were detected. A 15% change completely abolished the detection of GO categories. c. Some exemplary GO categories are shown for Group 1 (GP1) and Group 2 (GP2) genes. The two groups of genes are connected with completely different biological processes. GP1 genes are important for cell division and cell cycling, while GP2 genes mediate cell differentiation and functions that arise in specialized organs. The Y-axis indicates the normalized gene enrichment in GC3 codons; high negative values refer to genes containing high levels of AU3 codons. d. Difference in codon composition between the Group 1 genes (GP1), and the Group 2 (GP2) genes. The GP1 genes are rich in AU3 codons while GP2 genes are rich in GC3 codons. The segmented line indicates the average percentage of GC3 codons in all coding mRNAs of humans