In fboulnois/made: Microarray Analysis of Differential Expression

Introduction to the made pipeline

The made package provides tools for the reproducible analysis of oligonucleotide arrays using a simple, clear, and consistent interface.

This vignette will guide new users on using the features of the made pipeline for microarray analysis. The pipeline involves a few discrete steps, including generating the configuration and group file for the experiment, normalizing the microarray samples, performing gene level summarization, and generating the report.

All made experiments require a configuration file, or config. Preparing the config is covered in the "Defining the configuration and group file" section. Multiple entry points are then available depending on the input data. If the data available is the raw data, new users should consult the "Transforming raw data into an expression set" section. This converts the raw data into a suitable form for R known as an expression set (or eset), which describes the microarray experiment at the level of probes. Otherwise, if the user has already transformed their microarray data into an expression set, they should consult the "Summarizing an expression set" section.

Users should begin by loading the package in R using library(made).

Defining the configuration and group file

The first step for any made experiment is to create a configuration file, or config, which can be used by the various functions in the made package. The config consists of a set of desired options which can be used to reproducibly analyze a set of oligonucleotide microarrays. The config is generated using the write.yaml.config function, which has the following signature:

write.yaml.config(analysisDir, groupBy = NULL, ...)

The analysisDir parameter describes the directory in which the analysis will be performed.

The groupBy parameter must be one of "dirs", "files", or "eset", and describes the relationship between the microarray samples and groups. Groups will be used to compare different types of samples to each other. For example, one might have two groups, one for controls and one for disease. Grouping by "dirs" indicates that the analysis directory contains folders of samples and the folders themselves will be used to group the samples. Grouping by "files" indicates that the analysis directory contains a number of samples which will each be individually assigned to a group. Grouping by "eset" indicates that the user already has an expression set (eset) that they want to use in the analysis.

The ... parameter can be used to specify additional options, such as the statistical significance threshold, whether or not intermediate outputs should be saved, and whether a final report should be generated. It can be also used to define the groups programmatically, otherwise the user will need to modify the group file after the configuration has been created and assign the groups manually given the groupBy option they previously chose. A warning will be displayed to the user if this is the case.

More information regarding the parameters to the write.yaml.config function can be obtained by typing ?made::write.yaml.config in R.

Transforming raw data into an expression set

Once the config has been generated, raw data can be transformed into an expression set. The function to do this is ma.normalize, and accepts the config as a parameter to the function. The raw data is transformed into an expression set using the desired configuration options and returned from the function.

There are also other ways of converting raw data into an expression set. For example, it is possible to obtain normalized expression sets directly from the Gene Expression Omnibus (GEO). To do this, the GEOquery package needs to be installed. Once a user has identified a dataset of interest, they can use the getGEO(gse) function in the GEOquery package to download a list of all expression sets associated to the dataset.

Summarizing an expression set

Once the raw data has been transformed into an expression set, it can be summarized using the ma.summarize function. This function requires both the the config and the eset as input parameters. It outputs a summary of the experiment for each group comparison which includes the differentially expressed genes between each group and the linear models used to generate the summary. Additionally, if the GOstats and ReactomePA packages are installed, the summary also includes biological terms and pathways associated to the experiment.

fboulnois/made documentation built on May 16, 2019, 12:01 p.m.