Isochores, GC3 and mutation biases in the human genome

Abstract

In this work we re-examined the hypothesis that the variation in GC content in the human genome is due to different regional mutational biases. For this purpose we inferred the mutational pattern by using mutation databases that are available for many genes associated with human genetic diseases. The assumption of this approach is that such mutations reflect the actual frequency distribution of mutations as they arise in the population. Four classes of genes, classified according to their GC(3) level, were included in this study: GC(3)-poor genes (GC(3)<45%), genes with intermediate GC(3) content (45%<GC(3)<60%), GC(3)-rich genes (60%<GC(3)<75%) and very GC(3)-rich genes (GC(3)>75%). Our results show that most genes are under AT mutational biases, with very little variation compared to the expectations of neutral GC level. It is noteworthy that the mutational patterns in the GC(3)-rich genes do not appear to account for their GC(3)-richness. Instead, GC(3)-rich and very GC(3)-rich genes exhibit patterns of mutations that yield expectations of neutral GC(3) content that are much lower than their actual GC(3).