Exercise 11D: Comparing groups defined by two grouping variables in Prism

Go to parent GraphPad Prism statistical analyses

I measured the activity of the enzyme mannose-6-phosphate isomerase (MPI) in mice.
There are three genotypes at the mouse locus for MPI:

• Mpi^ff = 1
• Mpi^fs = 2
• Mpi^ss = 3

Since Prism only accepts numerical data the genotypes are represented by numbers.

I want to know whether the genotypes have different MPI activity. Because I don't know if sex affects MPI activity, I also recorded the sex of the mice (also represented by numbers: male = 0; female = 1). The data (in ΔO.D. units/sec/mg dry weight) are available as a csv file.

Open the file in WordPad. The data are arranged in a column table.

Open the file in Prism.
Create a new column table Select to Enter replicate values, stacked into columns Click Create Import the file in the table. The comma is used as a decimal separator.

Is the file in the correct format for analysis ?
No, when there are two grouping variables, Prism expects a grouped table. All two-factor analyses are linked to grouped tables.

Before you can create a grouped table, you have to choose which grouping variable to put in the rows and which in the columns, we'll put sex in the columns and genotype in the rows. So we create a table with three rows and two columns with 6 subcolumns each.

Sort the rows according to sex.

For sorting, you need an XY-column with the values that you want to base the sorting on in the X-column. Click the Change table type button in the upper left corner of the data table: In the Data table section of the Format Data Table window: set Kind of table to XY Click OK Select the sex column, right click and select Cut Select the X column, right click and select Paste Select the former sex column, right click and select Delete Sort the rows: you can find a detailed description on how to sort based on X-values in the section on sorting rows

Now you can create each of the three rows of the grouped table: each row has to contain the values of one specific genotype, first the six values for the males, then the six values for the females.

Create a table containing the data of the ff genotype.

Selecting specific rows (the ones with the ff genotype) from a table is done by pruning. For pruning you need an XY-column with the values that you want to base the selection on in the X-column. So instead of the sex column, you need the genotype column as the X-column. This means that you first have to switch back to a column table. To change an XY- into a column table Prism simply deletes the X-column and keeps the Y columns. Of course we want to keep the sex column so first copy the sex column and paste the values in the ID column (we don't need the IDs). Now we can change the table type back to a column table. Flip back to an XY-table, this time using the genotype column as the X-column. To prune the rows: Click the Analyze button in the Analysis section of the top toolbar: In the Transform, Normalize... section of the Analyze Data window select Prune rows Click OK In the X range to keep section set to to 1.0 to select the values of the ff genotype Click OK

Do the same for the fs and the ss genotype.

Store the data in a grouped table.

Create a new grouped table with 6 replicate values in side-by-side subcolumns. To insert the values: Go to the first pruned table (of the ff genotype) and select the values in the activity column, right click and select Copy Go to the new grouped table, select the first cell of the first row, right click and select Paste Transpose Do the same for the other two rows. Add column and row names. Remember: these can be text, they do no longer need to be numerical. Give a meaningfull name to the table.

Check whether genotype or sex influences MPI activity.

Analyze the data using a Two-way ANOVA (or mixed model) in the Grouped analyses section The RM Design tab: the data is independent: different mice are used for each measurement The Factor names tab you have to provide names for the factors The results indicate that the interaction term and the effect of sex are non-significant: However, genotype does influence the activity of the enzyme (effect of genotype is significant), albeit not in a sex-specific way (interaction is not significant).

Do all genotypes have different MPI activity ?
Redo the analysis. This time go to the Multiple comparisons tab and select to Compare row means (main row effect). The results indicate that only fs and ss are significantly different.

Again this exercise shows that often the most difficult step in Prism is getting your data in the correct format.