Skip to main content

Table 1 Comparison of Three Methods for Estimating Circadian Period in Locomotor Activity1

From: Signal analysis of behavioral and molecular cycles

Genotype n Autocorrelation2 MESA3 X2-Periodogram4
wild type(a) 18 24.5 ± 0.1 24.7 ± 0.2 24.4 ± 0.1
cyc01/+ 30 24.5 ± 0.1 24.5 ± 0.1 24.4 ± 0.0
cyc02/+ 20 24.7 ± 0.1 24.3 ± 0.1 24.7 ± 0.1
cyc-deletion/+ 20 24.6 ± 0.1 25.0 ± 0.2 24.6 ± 0.2
wild type(b) 24 24.3 ± 0.1 24.5 ± 0.2 24.4 ± 0.1
  1. 1. The data tabulated here come from experiments to analyze effects of mutations at the cyc locus on locomotor activity rhythms. After 5 7 days of entrainment to a light-dark cycle (LD 12:12), the endogenous period of the locomotor activity rhythm was evaluated based on 5 days in DD for the wild type(a), cyc01/+, and cyc02/+, or 6 in DD days for the remainder. The number of flies used for each estimate is given under the heading 'n.' The average estimate of circadian period ± the standard error of the mean is provided for each genotype (the rows) and each analytic method (the columns). Each of the 3 methods was applied to the same data set for comparison. cyc-deletion is a deficiency strain Df(3L) kto2, obtained from the Bloomington Stock Center.
  2. 2,3,4. Autocorrelation, MESA, and the X2-Periodogram provide numerical estimates of periodicity using different statistical approaches (see text). Non-parametric analysis by Wilcoxon's test indicates that the respective estimates produced by each method were not distinguishable: 29 of the 30 comparisons (10 comparisons for each method) revealed no significant differences between genotypes for any of the 3 methods (P > 0.01); the one exception was in the difference between cyc01/+ and cyc02/+ as evaluated by the chi-squared periodogram (P < 0.0001), however this difference of 0.2 hours is not considered reliable based on the limits of resolution in these studies (see text). In addition, Spearman's rank test was performed to evaluate pairwise comparisons between the three methods based on genotype. No consistent relationship between any of the methods is revealed by this test: significant correlations (p < 0.01) between the methods are evident in only 3 of the 15 comparisons. The association between autocorrelation and the chi-squared periodogram was significantly correlated for cyc02/+(p = 0.01) and cyc-deletion/+ (p < 0.0001). In addition, the association between MESA and the chi-squared periodogram was significantly correlated for cyc-deletion/+ (p = 0.0005).