Statistical analysis

All graphs and analyses were completed in R version 3.6.3 (https://www.r-project.org/) using R Studio version 1.3.1073 (https://www.rstudio.com/products/rstudio/). Custom R scripts were written and used for all statistical analyses. Linear regression was completed using lm() and included covariates as described. All mixed effect models were generated using lmer() within the lme4 package version 1.1-23 and included fixed and random covariates as described. For all box plots, the boxes represent the upper and lower quartiles and the whiskers represent variability; the median is illustrated as the black line within the box. Differences were considered significant at P<0.05. Additional details not already reported in the Results section are provided below.

A linear mixed model was used to obtain the variances explained by the between- and within-fish terms. The model was defined as: ECG Trait∼(1|fish identification variable)+(1|date of trace collection), where the first random effect is the between-fish term and the second is the within-fish term. The total variance explained was calculated by taking the sum of the variance explained by the between-fish term and the within-fish term, in addition to those of the residuals. The percentage of the total variance explained by each term was then calculated as 100 multiplied by the variance explained by the term divided by the total variance. The ratio was then calculated as the ratio of the percentage of the total variance explained by the between-fish term to the percentage of the total variance explained by the within-fish term. Variance-explained values <5×10⁻⁴ were rounded to zero before calculating the percentages.

To generate the null model in which there are no differences in the variance explained by the between- and within-fish terms, we permuted the fish identification variable such that the 42 traces were randomly assigned to different fish across the 4 days of recording. We then repeated the above analysis to obtain the percentage of the total variance explained for the between-fish and within-fish terms, in addition to the ratio.

The R² values were determined using a linear regression between the ECG metric calculated by zERG and the corresponding metric calculated using the ‘ECG Analysis’ module in LabChart.

After residuals were obtained from a linear mixed model adjusting the ECG trait for age, recording location and date of trace collection, model selection based on AIC values using a forward regression approach was completed using step().

For each independent experiment, traces were recorded in one recording session, fish were all the same age and the recordings occurred in the same location. Therefore, we did not adjust for date of ECG recording, age or recording location. The P-values were obtained from a linear regression adjusting for sex and weight; the QT interval was additionally adjusted for heart rate.

To account for differences in recording sessions conducted with fish from the kcnh6a^s290 line, we used a linear mixed model to analyze measurements obtained from the traces, with recording date added as a random effect and with age and location of ECG recording added as fixed effects. Additional covariates included sex, weight and heart rate as determined through model selection and correlation analyses for each respective ECG trait.

A linear mixed model was first used to adjust measurements of all ECG traits and obtain residuals; covariates included recording date (as a random effect), age of fish and location of ECG recording as fixed effects. Additional fixed effect covariates included sex, weight and heart rate (for QT interval only) as determined through model selection and correlation analyses for each respective ECG trait. Power calculations were then performed using power.t.test() after specifying the standard deviations of the residuals for each trait, the power, type (‘two.sample’), alternative (‘two.sided’) and delta (defined as the difference in means such that the percentage difference in means will be equivalent to the percentage defined in Table 2).