Process Description

Estimate Builder

The Estimate Builder process provides you with a quick and relatively easy method of constructing customized linear hypothesis tests in the form of SAS ESTIMATE statements. These statements are used in both ANOVA and mixed model analyses to test an arbitrary set of linear hypotheses regarding the relative importance of different combinations of fixed effects parameters. Statements are saved to a file that can subsequently be specified in the ANOVA or Mixed Model Analysis processes.

What do I need?

One data set, an Experimental Design Data Set (EDDS), is needed to run this process. This required data set tells how the experiment was performed, providing information about the columns of the primary experimental data. The drosophilaaging_exp.sas7bdat EDDS, associated with the Drosophila aging experiment from (Jin, et al., 2001) and described in Sample Case Studies, is used in the example that follows. This data set is shown below.

This data set is located in the Sample Data folder.

For detailed information about the files and data sets used or created by JMP Genomics software, see Files and Data Sets.

Example

This example uses the drosophilaaging_exp.sas7bdat EDDS shown above. Complete the following to generate a file of ESTIMATE statements:

8      Click to select the Experimental Design Data Set (EDDS).
8      Navigate into the SampleData \Microarray\Scanalyze Drosophila\ directory.
8      Select the drosophilaaging_exp.sas7bdat file and click .

The drosophilaaging_norm.sas7bdat file has been selected. If you click , the data set opens.

All of the data set variables are displayed in the Available Variables field. In this experiment, Line, Sex, Age, Array, and Channel all function as class variables.

8      Examine the list of available variables. Highlight Line, Sex, Age, Array, and Channel.
8      Click to add Line, Sex, Age, Array, and Channel to the Class Variables field.

The Estimate Builder focuses on the effects that different fixed factors, both individually and in combination, have on gene expression. To specify individual and potentially interacting factors, complete the following steps:

8      Type line sex age to allow for the investigation of the individual main effects of line, sex, and age. Leave a space between each effect and include no delimiters.
8      Type line*sex line*age sex*age, after the main effects, to allow for the investigation of the interaction of line with sex, line with age, and sex with age. Leave a space between each pair of interacting effects and include no delimiters.
8      Type channel to allow for the investigation of a dye main effect.
8      Type drosophila_estimates as the output filename.

To specify an output folder, complete the following steps:

8      Click .
8      Navigate to the folder where you want your results placed.
8      Click .

The completed dialog should appear as shown below.

8      Click to generate a second dialog window.

Note: The output file generated above is initially empty. It must be populated with estimate statements. Running Estimate Builder/Compare Means dynamically generates the Estimate Builder GUI window. This window lists all of the levels of the fixed effects specified in the dialog. Use the Estimate Builder GUI window to build specific ESTIMATE statements.

The Estimate Builder GUI window appears, as shown below.

Examine the Estimate Builder GUI window. Note that each of the main and interaction fixed effects, with all associated levels, is listed in the window. This window is used to specify the individual null hypotheses to be tested using anova or mixed model analysis. To create an ESTIMATE statement for the effect of sex on expression, complete the following step.

8      Click next to FEM to set its value to +1.
8      Click next to MAL to set its value to -1.

The settings for Sex should appear as shown below.

8      Type Sex in the Label for this Estimate field.
8      Click to generate the estimate statement.

An ESTIMATE statement appears as shown below.

Multiple estimate statements can be constructed and written to the same output file. To add a second estimate statement, to investigate the interaction of sex and age (Sex*Age), complete the following steps:

8      Click to rest all of the values for the fixed effects to zero.
8      Click next to FEM WK1 to set its value to +1.
8      Click next to FEM WK6 to set its value to -1.
8      Click next to MAL WK1 to set its value to +1.

Note: The value next to MAL WK1 does not go to +1, but instead goes to +0.5. If you examine the value next to FEM WK1, you should see that its value has changed to +0.5.

8      Click next to MAL WK6 to set its value to -1.

Note: The value next to MAL WK6 does not go to -1, but instead goes to -0.5. If you examine the value next to FEM WK6, you should see that its value has changed to -0.5.

The settings for Sex*Age should appear as shown below.

Note: The sum of the values equals to 0. Recall that in generating these estimate statements, we are setting the null hypothesis that there are no effects on expression. If the null is true, the sum is expected to be zero.

8      Type Sex*Age in the Label for the Estimate field.
8      Click Create an Estimate Statement Below to generate the estimate statement.

The ESTIMATE statements appear as shown below.

You must add the ESTIMATE statements to the output file created by the Estimate Builder.

8      Click to add the estimate statements to the output file.

The ESTIMATE statements are saved to the output file. The name and path to the output file is shown in a JMP Genomics Message window, as shown below.

Output/Results

The output of this process is a file containing the ESTIMATE statements. This file can be used in subsequent analyses (for example, ANOVA or Mixed Model Analysis), of the experimental data.