Fig. 2

AT-rich isochores in the human genome contain tandem arrays of genes with outward-looking functions. A, B UCSC Genome Browser screenshots showing GC% trajectory [65], CpG islands [66], our isochore calls, and simplified gene models. Gene models are colored according to k-means clusters described below. Gene name prefix diversity (Shannon’s H) is shown for the highlighted isochores. C Relationship between isochore GC% and gene contents. Isochores shown in A, B are highlighted. Colors show how we binned isochores into deciles according to GC%. Dashed line shows the mean GC content of the human genome (41%). D Boxplots representing gene name diversity (Shannon’s H) of gene-rich isochores across GC% deciles shown in C. Red points indicate the mean prefix diversity of the decile; black lines show medians. Here and throughout, most groups are statistically different from one another, except for adjoining groups; full statistical comparisons are presented in Additional file 4. E Comparison of number of genes in an isochore versus number of different gene name prefixes represented in that isochore, for isochores of at least five genes. Each dot is an isochore; isochores falling below the trend line contain clusters of genes with the same prefix. Isochores are labeled by the most common gene name prefix in that isochore. F Gene families with at least four members in isochores < 40% GC. Twenty-five percent of all genes are in isochores less than 40% GC (black line). Red bars depict proportions of genes with that prefix that are located in < 40% GC isochores. Gene family prefixes shown in black text are enriched in AT-rich isochores, while those shown in gray text (e.g., KRTAP, TRIM, SCN) have multiple family members in AT-rich isochores but are not enriched there. Functions of these gene families are marked at left, and total number of genes with that prefix in the MANE set are shown at right