library(dplyr)
library(biobroom)
mthd <- ifelse(x$method == "ls", "least squares", "robust regression")

The limma R/Bioconductor package can be used to find differentially expressed genes between groups of samples in an experiment using microarray or RNA-Sequencing data. The function lmFit() is used to fit a linear model for each of the genes including residual errors using r mthd for estimation. The function eBayes() is used to calculate the likelihood that a residual error would be seen by chance, or that a given is differentially expressed.

In this analysis, there were r dim(x$coefficients)[1] genes assessed for differential expression between r dim(x$coefficients)[2] groups (after removing for rows with expression levels all equal to 0). The top 10 genes that appear to be the mostly differentially expressed are

knitr::kable(topTable(x))

A volcano plot combines the measure of statistical significance with the magnitude of change in differential expression for each gene. The plots compares the log-fold changes versus log-odds of differential expression and is useful to identify large differences.

limma::volcanoplot(fit, main = "Volcano plot", coef = 2, highlight = 10)

The MAplot is a plot of the M-values (log-ratios) vs the A-values (average intensities).

limma::plotMA(fit, main = "MA plot", coef = 2)

A histogram of the p-values

hist(tidy(x)$p.value, main = "Histogram of p-values\nfor differential expression", xlab = "p-value")


hilaryparker/explainr documentation built on May 17, 2019, 3:58 p.m.