limmas

Limmas is the plural form used to refer to multiple analyses or results produced with LIMMA, the Linear Models for Microarray Data framework and Bioconductor software package. LIMMA provides statistical methods for detecting differential expression in microarray experiments and, more recently, RNA sequencing data. The approach was developed by Gordon K. Smyth and colleagues in the early 2000s and has since become a standard tool in transcriptomics.

The core idea of LIMMA is to fit linear models to expression data for each gene and then apply empirical Bayes moderation to the estimated gene variances. This borrowing of information across genes improves the accuracy of variance estimates, especially in experiments with small sample sizes. The package also supports a voom transformation, which converts RNA-seq counts into log-counts per million with a mean-variance relationship that can be analyzed by the same linear-model machinery. This allows LIMMA to be used for both microarray data and modern sequencing data within a consistent framework.

Typical workflows with LIMMA begin with data preprocessing, model fitting with lmFit, and specification of contrasts with contrasts.fit. Differential expression is then assessed using eBayes, which produces moderated test statistics and p-values. Researchers commonly extract results with topTable and may summarize multiple contrasts with decideTests. Limmas are widely used in genomics research due to their flexibility for complex experimental designs and robust performance in small-sample settings.

Note on usage: while LIMMA is widely recognized as an acronym, the plural limmas is not a formal term in the literature and may be used informally to denote multiple LIMMA analyses or results. See also LIMMA, Bioconductor, and differential expression analysis.