Fig. 3

Genes with outward-looking functions have high local AT content. A GC content trajectory for human protein-coding genes in the MANE set. Genes were subdivided by iterative k-means clustering. At top, the average GC content trajectory for each k-means cluster is shown as a line graph. At bottom, each gene is a row and GC content across the transcriptional unit and flanking regions is depicted from red (high AT) to blue (high GC). Rainbow colors assigned to each k-means cluster here will be used throughout. B Relationship between k-means cluster assignment and home isochore GC% for each gene in the MANE set. Red lines depict medians. C GO term distribution by promoter GC content for genes in the MANE set. Genes with immune, barrier, chemosensory, and xenobiotic functions have AT-skewed promoters. Genes with developmental and intracellular functions have GC-skewed promoters. Gray shading shows promoter GC content distribution of the whole MANE set. D Gene name prefixes enriched in cluster 3.3. Families shown have at least four members in cluster 3.3; proportion of the family located in this cluster is depicted in brown bars. Less than 10% of genes are in cluster 3.3 (“all”). E Gene prefix diversity (Shannon’s H) is lowest in cluster 3.3. F Distribution of promoter, intron, flanking sequence, and coding sequence GC% across genes which do not have paralogous neighbors in their home isochore (gray, genes with 0 tandem neighbors), genes which exist in small local duplications (red, genes with 1–4 tandem neighbors), and genes which exist in large multigene arrays (brown, genes with 5 or more tandem neighbors). A subset of genes without introns appears as 0’s