exprso (version 0.1.7)

compare: Compare ExprsArray Objects

Description

This method compares the values of all ExprsArray annotations across a specified annotation term for up to two ExprsArray objects. Depending on the composition of each annotation, compare will perform either a chi-squared test or an ANOVA test.

Usage

compare(object, array.valid = NULL, colBy = "defineCase", cutoff = 0.05)
"compare"(object, array.valid = NULL, colBy = "defineCase", cutoff = 0.05)

Arguments

object

The ExprsArray object used when comparing annotations.

array.valid

A second ExprsArray object used when comparing annotations. Optional. Exclude with array.valid = NULL.

colBy

A character string. The annotation column against which to compare all other annotation terms (i.e., to test as the independent variable).

cutoff

A numeric scalar. The p-value cutoff that determines when the annotation test returns a TRUE result

Value

A list of three logical vectors. The first and second elements of the list correspond to "internal" comparisons for the two provided ExprsArray objects, respectively. The third element of the list corresponds to comparisons made between the provided objects.

Methods (by class)

• ExprsArray: Method to compare ExprsArray objects.

Details

This method performs two kinds of comparisons. First, it tests all annotation variables against the annotation supplied by the colBy argument for each provided ExprsArray object. In other words, the colBy argument determines which annotation to use as the independent variable for "internal" comparisons. Second, it tests all annotation variables between the provided ExprsArray objects. Providing array.valid = NULL will skip the between comparisons.

This method will test annotations using either a chi-squared test or an ANOVA test depending on the class of the values stored by the tested column. The presence of a "character" or "factor" in the tested column will trigger a chi-squared test. As such, this method requires the user to select a colBy annotation that contains categorical data (i.e., to use as the independent variable).

We anticipate that this method will serve as a useful adjunct to modCluster. However, it may also help in quickly determining whether the data split has yielded comparable training and test sets in terms of the annotations included in @annot.