Fig. 1. Correlation function Γ(d) of the GC% series with 2^k GC values (k=15 : dots, top; and k=17 : lines, bottom), obtained from all 24 human chromosomes (22 autosomal and 2 sex chromosomes). The x -axis, in a logarithmic scale, is the distance d converted to units of Mb (10⁶ bases).

Fig. 2. Correlation Γ_w(d), at the distance d≈ 1 Mb, as a function of the chromosome-specific window size w (in a double logarithmic plot). Each point represents one human chromosome. The window size w is the chromosome length divided by 2¹⁷ .

Fig. 3. Correlation function Γ(d) of the GC% series for the DNA sequence of human chromosome 21. GC% is calculated at the window size of 358 bases, which is 1/2¹⁷ of the total chromosome length. The distances of 0.5, 1, 1.5, and 2 Mb are marked by long vertical lines, and the spacing of 100 kb is marked by short vertical lines.

Fig. 4. Correlation function Γ(d) of the GC% series for eight chromosomal regions of the DNA sequence of the human chromosome 21. The distances of 0.5, 1, 1.5, and 2 Mb are marked by vertical lines.

Fig. 5. GC% fluctuations of the last one-eighth segment of the DNA sequence of human chromosome 21 towards the q-term end. (a) GC% calculated for the window size of w=2.864  kb, which is 1/2¹⁴ of the whole chromosome length. A sinusoidal function with the period of 500 kb is superimposed on the plot to fit the periodic oscillation of GC%. (b) GC% calculated with interspersed repeats replaced by random values, then smoothed by means of running medians, using the S-PLUS subroutine smooth. (c) Locations of known genes as determined by protein sequences from SWISS-PROT, TrEMBL, and TrEMBL-NEW, and their corresponding mRNAs from GenBank, displayed separately each two DNA strand. (d) Locations of CpG islands (top) and isochores (bottom). For the isochore map, the GC% of individual isochores is indicated by the height of the horizontal bar.