4. Interpret Results

The next step is to interpret the results from the CNV PCA Search script and determine the number of principal components to use for the principal component analysis.

A. Determine Number of PCs with Slope Closest to One

• From the PCA Search Results spreadsheet, right-click the Slope column header and select Sort Ascending.
• Scroll to the number of components that has a Slope closest to one.

In this example, the slope is closest to one for 31 principal components.

Figure 4a. PCA Search Results displaying slope closest to one and the smallest -log10(F) value.

Note

A slope of one indicates that the observed values are in line with the expected values, thus indicating that the observed p-values that are not significant are no more or no less significant than expected. You can also look for the number of components that has the smallest -log10(F) value. In most cases, this will be consistent with the number that has a slope closest to one. In this example however, there are T-cell artifacts that cause there to be highly significant results. These values are going to cause the optimal number of principal components determined by the slope (31) and by the F statistic (38) to be different.