Clay Oliver, Bernardi Giorgio
Mol Biol Evol. 2005 Dec;22(12):2315-7. doi: 10.1093/molbev/msi231. Epub 2005 Aug 10.
In a recent paper in these pages, Cohen et al. search for isochores in the human genome, based on a system of attributes that they assign to isochores. The putative isochores that they find and choose for presentation are almost all below 45% GC and cover only about 41% of the genome. Closer inspection reveals that the authors' methodology systematically loses GC-rich isochores because it does not anticipate the considerable fluctuations and corresponding long-range correlations that characterize mammalian DNA and that are highest in GC-rich DNA. Thus, they over-fragment GC-rich isochores (and also many GC-poor isochores) beyond recognition.
在本刊近期发表的一篇论文中，科恩等人基于他们赋予等密度区的一系列属性，在人类基因组中寻找等密度区。他们所发现并挑选用于展示的假定等密度区，几乎全部低于45%的GC含量，仅覆盖了约41%的基因组。进一步审视会发现，作者的方法系统性地遗漏了富含GC的等密度区，因为该方法没有考虑到哺乳动物DNA所具有的显著波动和相应的长程相关性，而这些在富含GC的DNA中最为明显。因此，他们过度分割了富含GC的等密度区（以及许多GC含量低的等密度区），使其无法被识别。
