Seven unique frequency profiles for scoring vigilance states in preclinical electrophysiological data

Figures

• 

  Figure 1. Frequency profiles of all states. (A) The profiles of all states used in scoring, both with the profile from FFT (AccuSleep) (blue) and the power spectral density using Welch’s (turquoise). (B) The coefficient of variance (CV) from the FFT, for the three awake states (a1; blue, a2; yellow, a3; red) along with the total awake (black). A.u.—arbitrary unit.

• 

  Figure 2. (A) Summary of labels from manual scoring and semi-automatic scoring. (B) Mean α exponent. Black: manual scoring, turquois: SAS. The two curves differ significantly at three frequencies out of 65. This happens in Asterisk where the CI95 (Bonferroni-adjusted) of the difference wave does not include 0. This happens at the alpha band frequency.

• 

  Figure 3. Epoch lengths primarily showed an effect on the less frequent states, such as theta dominated awake, alpha dominated awake (resting state), and REM sleep. To have comparable amounts of data, we used the α exponent of the total awake state data and not the three substates separately. Blue; 0.5 s, yellow; 1 s, red; 5 s.

• 

  Figure 4. Proportion of vigilance states varies with genotype. (A) Summary of label distribution for three genotypes [wild-type (WT), heterozygous (Het), and homozygous (Hom)] at two time points [8 weeks (w8) and 22 weeks (w22) of age] scored with 1 s epochs. (B) Correlation plot of variables: rose is Pearson’s correlation coefficient of 1 = positive correlation, green: ~ − 1 = negative correlation. Mint: 0, no relation between variables. Age showed barely any correlation but a significant relation to REM (p = 0.03). Genotype showed statistically significant relations to all states especially nREM1 (Pearson’s = −0.74, p < 0.0001). (C) Correlation of proportion of nREM1 and a1, slope = −1.1 for WT and Het, but not for Hom at the second time point slope = 0.64.

• 

  Figure 5. Hourly distribution of wake and sleep states over the course of 24 h, starting at 9 am on day 1 and ending at 9 am on day 2 (x-axis). The dark phase is indicated by the black frames. Wild-type (WT), heterozygous (Het), and homozygous KO (Hom) of Nrxn1 at (A) 8 weeks (young adult) and (B) 22 weeks of age (middle-aged). Statistical analysis of the development in a1 and nREM1 epochs is shown in Figure 6.

• 

  Figure 6. Cumulative distributions shown for (A) a1 and (B) nREM1 in all three genotypes studied at 8 and 22 weeks of age. The Hom group has both more a1 and less nREM1 than the other groups. The x-axes are the cumulated number of epochs; the y-axes are the proportion of data points (one step per hour). The square frames indicate the dark/active phase. The asterisks indicate when the Bonferroni-corrected CI95 of the difference curve to WT1 does not include 0, meaning a statistically significant difference. Red; WT time point 1 (8 weeks), dark green; Het 1, light blue; Hom 1, yellow; WT time point 2 (22 weeks), light green; Het 2, pink; Hom 2.