outliers: Helper functions for outlier detection and reporting in...
In grimbough/arrayQualityMetrics: Quality metrics report for microarray data sets

Description Usage Arguments Details Value Author(s)

View source: R/outlier.R

For an overview of outlier detection, please see the corresponding section in the vignette Advanced topics: Customizing arrayQualityMetrics reports and programmatic processing of the output. These two functions are helper functions used by the different report generating functions, such as aqm.boxplot.

1 2	outliers(exprs, method = c("KS", "sum", "upperquartile")) boxplotOutliers(x, coef = 1.5)

`exprs`	A matrix whose columns correspond to arrays, rows to the array features.
`method`	A character string specifying the summary statistic to be used for each column of `exprs`. See Details.
`x`	A vector of real numbers.
`coef`	A number is called an outlier if it is larger than the upper hinge plus `coef` times the interquartile range. Upper hinge and interquartile range are computed by `fivenum`.

outliers: with argument method="KS", the function first computes for each column of exprs (i.e. for each array) the value of the ks.test test statistic between its distribution of intensities and the pooled distribution of intensities from all arrays. With "sum" and "upperquartile", it computes the sum or the 75 percent quantile. Subsequently, it calls boxplotOutliers on these values to identify the outlying arrays.

boxplotOutliers uses a criterion similar to that used in boxplot.stats to detect outliers in a set of real numbers. The main difference is that in boxplotOutliers, only the outliers to the right (i.e. extraordinarily large values) are detected.

For outliers, an object of class outlierDetection. For boxplotOutliers, a list with two elements: thresh, the threshold against which x was compared, and outliers, an integer vector of indices.

Wolfgang Huber

grimbough/arrayQualityMetrics documentation built on May 3, 2019, 4:06 p.m.