10, 100 kb and 1 Mb, respectively. Among all genomic features we examined, we found that local GC content is the most correlated factor with median signal intensities. In addition, for both samples, the magnitude of correlation between signal intensity and GC content increases for larger window sizes (Figure 2B–D and F–G). To further illustrate the correlation of the signal intensity values and the six genomic features in 1 Mb windows, we plotted the values along a representative chromosome for the two samples, and demonstrated the clear correspondence of peaks and troughs between signal intensity and GC content (Figure 3). Finally, we plotted the signal intensity values and GC content in 1 Mb windows in the entire genome, and demonstrated a consistent pattern of correspondence across all chromosomes (Supplementary Figure 1). Altogether, our analysis indicates that local GC content is the best predictor of the magnitude of local genomic waves among the features we evaluated. To distinguish the directionality of wavy patterns in different samples, we refer to waves that have positive correlation with GC as positive waves (Figure 2A) and those with negative correlation as negative waves (Figure 2E). Table 1.Correlation of wave with genomic featuresGenomic featuresSample 1Sample 210
