diff_expr {crossmeta}R Documentation

Differential expression analysis of esets.

Description

After selecting control and test samples for each contrast, surrogate variable analysis (sva) and differential expression analysis is performed.

Usage

diff_expr(esets, data_dir = getwd(), annot = "SYMBOL",
  prev_anals = list(NULL))

Arguments

esets

List of annotated esets. Created by load_raw.

data_dir

String specifying directory of GSE folders.

annot

String, column name in fData common to all esets. For duplicated values in this column, the row with the highest interquartile range across selected samples will be kept. If meta-analysis will follow, appropriate values are "SYMBOL" (default - for gene level analysis) or, if all esets are from the same platform, "PROBE" (for probe level analysis).

prev_anals

Previous result of diff_expr, which can be reloaded using load_diff. If present, previous selections, names, and pairs will be reused.

Details

The Samples tab is used to select control and test samples for each contrast. To do so: select rows for control samples, type a group name in the Control group name text input box and click the Add Group button. Repeat for test samples. While adding additional contrasts, a previous control group can be quickly reselected from the Previous selections dropdown box. After control and test samples have been added for all contrasts that you wish to include, click the Done button. Repeat for all GSEs.

Paired samples (e.g. the same subject before and after treatment) can be specified by selecting sample rows to pair and then clicking Pair Samples. The author does not usually specify paired samples and instead allows surrogate variable analysis to discover these inter-sample relationships from the data itself.

The Contrasts tab is used to view and delete contrasts that have already been added.

For each GSE, analysis results are saved in the corresponding GSE folder in data_dir that was created by get_raw. If analyses needs to be repeated, previous results can be reloaded with load_diff and supplied to the prev_anals parameter. In this case, previous selections, names, and pairs will be reused.

Value

List of named lists, one for each GSE. Each named list contains:

pdata

data.frame with phenotype data for selected samples. Columns treatment ('ctrl' or 'test'), group, and pairs are added based on user selections.

top_tables

List with results of topTable call (one per contrast). These results account for the effects of nuissance variables discovered by surrogate variable analysis.

ebayes_sv

Results of call to eBayes with surrogate variables included in the model matrix.

annot

Value of annot variable.

Examples

library(lydata)

# location of raw data
data_dir <- system.file("extdata", package = "lydata")

# gather GSE names
gse_names  <- c("GSE9601", "GSE15069", "GSE50841", "GSE34817", "GSE29689")

# load first eset
esets <- load_raw(gse_names[1], data_dir)

# run analysis
# anals <- diff_expr(esets, data_dir)

# re-run analysis on first eset
prev <- load_diff(gse_names[1], data_dir)
# anals <- diff_expr(esets[1], data_dir, prev_anals = prev)

[Package crossmeta version 1.6.0 Index]