Fig. 7
Association between disease and oscillating cytosine modifications. a Odds ratios of overlap between osc-modCs and differentially modified loci in various disease datasets estimated using Fisher’s exact test. Full circles mark log2 odds ratios, and whiskers represent 95% confidence intervals. b Scatterplot showing the association between the osc-modC amplitude and log-transformed odds ratio for type II diabetes risk. The black line indicates fitted least squares regression line with shaded gray area depicting 95% confidence interval. c Box plots of cytosine modification differences in simulated and a representative EWAS hit, cg10311104, from schizophrenia EWAS ([32]; Supplementary Fig. S5). Black dots represent outlier samples beyond the interquartile range. d Ten representative samples from osc-modC simulation in the “control” and the “patient” groups. Gray boxes represent the regular “office hours” (9 AM–5 PM) when samples are usually collected in a realistic clinical setting. Black curves represent the oscillation profiles for each sample, with red dots indicating a randomly selected sample collection time. CLL, chronic lymphocytic leukemia; C.I., confidence interval; modC, modified cytosine; CT, circadian time; CTRL, control; SCZ, schizophrenia; OBS, observation