Example 3

Effect of genotype on plasma glucose

View experiment diagram | Use this diagram as a template

This is a phenotyping experiment, designed to test whether the mice genotype (wild type or knock out) affects plasma glucose levels following a glucose challenge and whether the effect of genotype depends on the sex of the animal. Four groups of mice of all the possible combinations of sex and genotype are used. The experimental unit in this experiment is each animal; this is only indicated for group 1 but the system assumes that the experimental unit is the same for all groups.

The independent variables of interest here are sex and genotype. There is no allocation node on this diagram as the genotype and sex for each animal are randomised by Mendelian inheritance prior to the experiment. This is indicated in the properties of the variable nodes for sex and genotype.

Mice are fasted for 14h before the start of the experiment, then body weight is measured for all animals. For each mouse, at t=0 plasma glucose is measured. Immediately after, an injection of glucose is administered. Each mouse is individually coded so that the experimenter does not know which group the animals belong to, blinding is however only partial as the experimenter can see the sex of the animals when injecting them. Animals are processed in a random order so that the time of the day does not confound the effect of sex and genotype. At t=15 min, t=30, t=60 and t=120 min, all animals are measured again, in the same random order as before so that the interval between injection and measurements is similar for all animals.

Note that if measurements were spaced by a larger interval, for example one week instead of one hour, animals could be re-randomised for each measurement as re-ordering the animals would have limited impact on the intervals between measurements.

The analysis uses a summary measure – the area under the curve – to compare plasma glucose over 2h between the four groups. This analysis includes genotype and sex as factors of interest, the experimenter is interested in the overall effect of sex and genotype on blood pressure and plasma glucose between the groups rather than differences between time points, thus the independent variable ‘time of measurement’ is not included in the analysis but a summary measure is used instead. If the data fit parametric assumptions, it can be analysed using a factorial ANOVA (2 way ANOVA with interaction).

Group sizes are calculated based on the planned analysis method, to ensure that the experiment yields enough power to detect a difference between the groups if there is one. Power calculations for 2 way ANOVA are not available within the EDA so G*Power is used using the following settings and input parameters:

  • Test family: F tests
  • Statistical test: ANOVA: fixed effects, special, main effects and interactions
  • Type of power analysis: A priori: compute required sample size – given α, power and effect size
  • Effect size f: 0.4 (large effect by Cohen’s definition)
  • α err prob: 0.05 (this is the significance level)
  • Power (1-β err prob): 0.8 (this is the statistical power)
  • Numerator df: 1 (number of categories in each variable minus 1, both variables have 2 categories)
  • Number of groups: 4

The total sample size calculated is 52, which relates to 13 experimental units (13 animals) per group.

The outcome measure ‘body weight’ is not included in the analysis as it is measured to calculate the dose of the glucose challenge.

Keywords

Multiple animal characteristics | No randomisation | Data transformation | Multiple time points | Multiple outcome measures | Sex and genotype as factors of interest | Two-way ANOVA with interaction

References

This experiment is loosely based on the intraperitoneal glucose tolerance test (ipGTT) from the International Mouse Phenotyping Resource of Standardised Screens (IMPReSS)