RSEM (RNA-Seq by Expectation-Maximization)

Overview

RSEM uses an expectation-maximization algorithm to estimate transcript- and gene-level expression from RNA-seq data, explicitly modeling multi-mapping reads by distributing read counts probabilistically across isoforms. It outputs expected counts, TPM (transcripts per million), and FPKM/RPKM values. RSEM is typically used after STAR or HISAT2 alignment and is the standard quantification tool in TCGA and many cancer genomics pipelines.

Used by

  • Used to quantify transcript-level gene expression from STAR-aligned RNA-seq reads (hg38) for 28 metastatic neuroendocrine neoplasms (pog570_bcgsc_2020) in the BC Cancer POG WGTA pipeline; expression values fed into edgeR differential expression, consensus hierarchical clustering, t-SNE visualization, and VIPER master-regulator analyses PMID:24326773.
  • Used for RNA-seq quantification in bulk transcriptomics of 35 AAV-CRISPR-edited rat mammary tumors, enabling ANOVA-based identification of 1,579 differentially expressed genes and GSEA-based comparison with human endocrine therapy response datasets PMID:26437033
  • RSEM used for transcript-level quantification of RNA-seq data from 17 ACC tumors; confirmed NFIB overexpression vs normal tissue (p=0.002) independent of fusion status, and MYB overexpression in fusion-positive tumors PMID:26862087
  • MapSplice/RSEM pipeline used for mRNA quantification in the TCGA esophageal/stomach study of 164 oesophageal carcinomas and 359 gastric adenocarcinomas PMID:28052061.
  • Used to quantify transcript-level expression from RNA-seq data in AALE chr_3p-deleted cell experiments and in the TCGA pan-cancer expression dataset (Broad GDAC Firehose 2016_01_28 release) PMID:29622463

Notes

  • RSEM estimates are probabilistic; multi-mapping reads across highly similar isoforms are distributed by expectation, not assigned deterministically.
  • RSEM expected counts (not raw counts) are suitable input for edgeR/DESeq2 when passed as rounded integers; TPM is preferred for cross-sample comparisons.
  • RSEM requires pre-built reference indices (rsem-prepare-reference) that include transcript and genome FASTA plus annotation GTF.

Sources

This page was processed by crosslinker on 2026-05-09. - PMID:26437033

This page was processed by wiki-cli on 2026-05-14. - PMID:26862087

This page was processed by wiki-cli on 2026-05-14. - PMID:28052061

This page was processed by wiki-cli on 2026-05-14. - PMID:29622463

This page was processed by wiki-cli on 2026-05-15.