Widom J
Department of Biochemistry Molecular Biology, Cell Biology, Northwestern University, Evanston, IL 60208-3500, USA.
J Mol Biol. 1996 Jun 21;259(4):579-88. doi: 10.1006/jmbi.1996.0341.
Fourier transform techniques have been used to analyze the distributions of all ten independent DNA dinucleotide steps in two eukaryotic genomes and one prokaryotic genome, for periodicities of approximately 2 to 500 bp. The results reveal systematic deviations from random expectation for certain dinucleotide steps over this entire range of periodicities, together with striking peaks at certain spatial periodicities for particular dinucleotide steps. Several dinucleotides yield peaks at a periodicity of approximately 10.2 bp that are unique to the eukaryotic genomes. Certain members of this set of dinucleotide signals were previously identified as involved in nucleosome positioning, while others were previously unrecognized. In real-space, these dinucleotides are uncorrelated or even anticorrelated (relative to random expectation) at distances of 10 and 11 bp, despite having greater than random spectral power at the corresponding periodicity. Real-space correlations of these dinucleotides at distances of 10 and 11 bp are suppressed by another spectral component, a 3 bp periodicity attributed to codons, which has a local minimum probability at approximately 10.5 bp. When the two eukaryotic genomes are encoded for the signal "AA or TT", the peak at approximately 10.2 bp periodicity is strengthened, whereas for the prokaryotic genome such a peak remains absent. For the Caenorhabditis elegans genome, this peak becomes the dominant feature in the transform, surpassing a peak owing to the existence of codons in both height and integrated intensity. These results suggest that the requirements of chromosome structure place significant constraints on eukaryotic genome organization; they reveal additional signals that may be related to nucleosome positioning; and they reveal a wealth of additional new non-random aspects of genome sequence organization.
傅里叶变换技术已被用于分析两个真核生物基因组和一个原核生物基因组中所有十个独立的DNA二核苷酸步移的分布情况，分析周期范围约为2至500碱基对。结果显示，在整个周期范围内，某些二核苷酸步移与随机预期存在系统性偏差，同时特定二核苷酸步移在某些空间周期处出现显著峰值。几种二核苷酸在约10.2碱基对的周期处产生峰值，这是真核生物基因组所特有的。这组二核苷酸信号中的某些成员先前已被确定与核小体定位有关，而其他成员此前未被识别。在实空间中，尽管这些二核苷酸在相应周期处具有大于随机的谱功率，但在10和11碱基对的距离处它们是不相关的，甚至是反相关的（相对于随机预期）。这些二核苷酸在10和11碱基对距离处的实空间相关性被另一个谱成分抑制，即归因于密码子的3碱基对周期，该周期在约10.5碱基对处具有局部最小概率。当两个真核生物基因组编码信号“AA或TT”时，约10.2碱基对周期处的峰值会增强，而原核生物基因组中则不存在这样的峰值。对于秀丽隐杆线虫基因组，这个峰值在变换中成为主导特征，在高度和积分强度上都超过了由于密码子存在而产生的峰值。这些结果表明，染色体结构的要求对真核生物基因组组织施加了重大限制；它们揭示了可能与核小体定位相关的额外信号；并且它们揭示了基因组序列组织中大量新的非随机方面。
