25.7.1 Introduction

The LogR Association Tests and PCA Window allows you to run association tests and Principal Component Analysis (PCA) directly on a LogR dsf file. These features can be run simultaneously or separately. This window has the same type of output as the Genetic Association Test Window (see 18.8.7).

25.7.2 Data Requirements

Association testing requires both a LogR dsf file (see 25.2) as well as a spreadsheet of phenotypes. If only PCA correction is needed, only the LogR dsf file is required. The association testing and PCA correction will only be run on patients (rows) that are activated in the phenotype spreadsheet and are also in the LogR dsf file. Note the spreadsheet phenotype must have row labels.

25.7.3 Available Tabs

There are three tabs in the LogR Association Tests and PCA window:

• Input. This tab contains all of the main options for what association tests are to be used and whether PCA correction is to be used.
• Principal Component Analysis Parameters. This tab contains parameters for PCA correction, including the number of principal components, outputting the eigenvalues to a spreadsheet, and whether outliers are to be removed.
• Exclude SNPs. This tab is used to exclude certain SNPs from the analysis.

25.7.4 Input Tab

---

---

The input tab contains three main sections:

• Input File. You must select a valid LogR dsf file to do association testing/PCA correction on. You can also choose specific chromosomes to include.
• Association Tests. This is where you can select which association tests to run and which phenotype spreadsheet to use.
• PCA Correction. This is where you can enable PCA correction, outputting the corrected data, and whether the dependent variables are to be PCA correction also (only applies if association tests are done also).

25.7.5 Description of Association Testing

Association Testing using this window uses an Additive model (see 18.2.3) for all processing.

• For more information on the Correlation/Trend Test, see 18.3.1
• T-test can only be run when the phenotype dependent trait is boolean (or can be transformed into a boolean choice - such as a gender trait). For more information, see 26.22.9
• Regression can be run on either boolean or quantitative (continuous) traits. If a boolean trait is selected, Logistic Regression (see 26.16) is used. Otherwise, Linear Regression (see 26.7.2) is used.

Note: If you have association tests selected, you must select a Phenotype spreadsheet before you can run analysis.

25.7.6 PCA Correction

If you have PCA correction selected, you have the option to output the corrected LogRs to a DSF. This allows you to import and segment the corrected data using the Copy Number Segmentation window (or import the LogR values into HelixTree), or use it as the input dsf in this window without having to re-run PCA correction each time.

If you have both PCA correction and association tests selected, you can choose to PCA correct the dependent variable or not. If you are using both uncorrected and corrected results, each test will have two sets of columns - one set for the corrected data and one for the uncorrected data.

Note: As the T-test only works with boolean dependent variables, using a corrected dependent does not change the output.

Note: For the corrected Regression results, be aware that a Linear Regression will be done whether the dependent variable was initially boolean or not.

25.7.7 Principal Component Analysis Parameters Tab

---

---

This tab contains options specifically related to Principal Component Analysis. See 18.6.1 for a description of these parameters. Note: As PCA correction in this case is of LogR data rather than genotype data, there is no normalization.

25.7.8 Exclude SNPs Tab

---

---

If desired, use the Browse button to open a CSV or text file containing SNPs to exclude from association testing and PCA correction. Use the viewer to preview the SNPs that will be excluded. To remove the list of SNPs, use the Clear List button.