Construct a PCAGinzu object

We begin principal component analysis (PCA) and interpretation of a given dataset by simply constructing one of the "PCAGinzu" class of objects, e.g.:

p = pcaGinzu.pcaGinzu(cho,verbose=True)

In the above example, we constructed a basic "pcaGinzu" class object given the Cho dataset. We also turned on an optional verbose flag so we could observe what the is happening during construction of the PCAGinzu object. Behind the scenes three things happened:

1. a RowPCAView is constructed- this is a "view" of the data after the PCA eigenvectors are calculated and the data has been rotated into the PCA space;
2. the genes (rows) at the extremes of each principal component are identified and labeled; and
3. the conditions (columns) in which those extreme genes (rows) are identified and labeled.

There are two different sub-types of PCAGinzu objects that one can instantiate, each providing a slightly different graphical view of the analyzed dataset. The "PCAGinzu" class is defined in compClust.iplot package (See iplot import command in section 2.1). It will create interactive plots using the IPlot API (which can be rendered in either Tk or static plots (for web pages)).

Alternatively there is also "pcaGinzuVisualizeMatplotlib" class within compClust.mlx.pcaGinzu module, which will render static, publication-quality plots using matplotlib. pcaGinzuVisualizeMatplotlib also has an additional batch processing function which can iterate over all of the principal components and create all the available PCAGinzu analysis output for each component.

No matter which version of PCAGinzu object we create, we need to pass in the dataset we want to analyze. The dataset is the only required argument to these constructors. The following subsections describe some of the pcaGinzu object's optional parameters that further control the PCA interpretation analysis.