- Unique molecular identifiers (UMIs) and cell barcodes are essential molecular components of many single-cell RNA sequencing (scRNA-seq) technologies. They make it possible to identify the cell from which an RNA molecule originated and distinguish individual RNA molecules from copies generated during amplification. By combining these two types of identifiers, researchers can measure gene expression at single-cell resolution, reduce amplification-related counting errors and investigate cellular diversity in complex tissues.
- A cell barcode is a short sequence of nucleotides incorporated into the sequencing library to identify the cell associated with a particular RNA molecule. In droplet-based single-cell RNA sequencing, individual cells are encapsulated in droplets together with beads carrying cell-specific barcode sequences. Each bead contains many copies of the same cell barcode, allowing RNA molecules captured from that cell to receive a common identifier. After reverse transcription and library preparation, sequencing reads containing the same cell barcode can be assigned to the same cell. This enables the reconstruction of gene-expression profiles for thousands or even millions of individual cells, depending on the platform and experimental design.
- Cell barcodes are particularly important in high-throughput droplet-based systems such as 10x Genomics Chromium, Drop-seq and inDrop. In these methods, cells are separated into individual droplets, and barcode-bearing beads introduce cell-specific identifiers during RNA capture. After the droplets are broken and their contents are pooled, all captured RNA molecules can be processed together because their barcode sequences preserve information about their cellular origin. Cell barcodes therefore allow researchers to combine material from many cells in a single sequencing library without losing the identity of individual cells.
- Unique molecular identifiers are short, usually random or semi-random nucleotide sequences attached to individual RNA-derived molecules during library preparation. Unlike cell barcodes, which identify the cell of origin, UMIs identify individual captured RNA molecules. A UMI is associated with a transcript-derived cDNA molecule before substantial amplification takes place. During sequencing-data analysis, reads that map to the same gene or transcript and share the same cell barcode and UMI can be grouped together. These reads are generally treated as PCR duplicates originating from the same original molecule rather than as independent RNA molecules.
- The distinction between cell barcodes and UMIs is fundamental. A cell barcode answers the question, “Which cell did this RNA molecule come from?” A UMI answers, “Which original RNA molecule does this sequencing read represent?” Both identifiers are required in many droplet-based workflows to reconstruct gene expression accurately. The combination of cell barcode, UMI and gene or transcript assignment provides the basis for generating a cell-by-gene expression matrix.
- PCR amplification is necessary in RNA sequencing because the amount of RNA captured from an individual cell is usually very small. However, amplification can produce many copies of the same original cDNA molecule. If every sequencing read were counted as an independent transcript, highly amplified molecules would be overrepresented and gene-expression estimates could become biased. UMIs help address this problem by allowing reads derived from the same original molecule to be collapsed into a single molecular count. This makes UMI-based quantification more reliable than simply counting sequencing reads, although it does not eliminate all technical errors.
- The molecular complexity of a UMI depends on its length and nucleotide composition. For a random UMI containing four possible nucleotides at each position, the theoretical number of possible sequences increases exponentially with length. For example, a 10-nucleotide random UMI can theoretically generate 4¹⁰, or 1,048,576, different sequences. Longer UMIs provide a larger identifier space and reduce the probability that two independent molecules receive the same UMI by chance. However, UMI collisions can still occur, particularly when many molecules are captured, and sequencing errors can make identical original UMIs appear different. Computational UMI-correction methods are therefore used to improve molecular counting.
- The theoretical number of possible UMIs can be expressed as:
- Number of possible UMIs = 4ᴸ
- Number of possible UMIs = 4^L
- Here, 4 represents the four possible nucleotide bases (A, T, G and C), and L represents the UMI length in nucleotides. For example, a UMI of length 10 has 4¹⁰ possible sequences. This is a theoretical sequence space; the actual probability of collisions depends on the number of molecules captured, the UMI design and the distribution of sequences.
- During single-cell RNA-seq library preparation, cell barcodes and UMIs are commonly introduced during reverse transcription. Polyadenylated RNA molecules are captured using oligo-dT primers, which bind to the poly(A) tails of mRNAs. The primers contain sequences for cell barcoding, UMI identification and subsequent library amplification. The captured RNA is reverse-transcribed into cDNA, after which the cDNA is amplified and converted into sequencing libraries. The exact arrangement and location of barcode and UMI sequences vary between platforms and protocols.
- In droplet-based systems, the cell barcode is generally shared by RNA molecules captured from the same cell, whereas each captured molecule receives its own UMI. In plate-based methods, cells are separated into individual wells, and the well identity can serve as the cell identifier. Some plate-based protocols also introduce molecular barcodes and UMIs. Smart-seq2, for example, is commonly used for full-length transcript analysis but does not use UMIs in its standard protocol. Smart-seq3 incorporates UMIs to support molecular quantification while retaining information about transcript structure.
- After sequencing, bioinformatics pipelines extract cell barcodes and UMIs from the appropriate read positions. Reads are aligned to a reference genome or transcriptome, and their gene or transcript assignments are determined. Reads are then grouped according to their cell barcode, gene assignment and UMI. Duplicate reads representing the same original molecule are collapsed, and the resulting molecular counts are used to construct a gene-expression matrix. Each row typically represents a cell, each column represents a gene, and each matrix entry represents the number of detected molecules assigned to that gene in that cell.
- Several computational tools support barcode processing and UMI-based quantification. Cell Ranger, developed for 10x Genomics data, performs barcode processing, read alignment, UMI counting and cell identification. STARsolo provides a flexible alternative for processing single-cell RNA-seq data using the STAR aligner. Alevin-fry supports efficient processing and quantification of single-cell RNA sequencing data. UMI-tools provides methods for extracting, grouping and correcting UMIs, including approaches for identifying likely sequencing errors and handling UMI collisions. These tools differ in their algorithms, supported protocols and assumptions, so the appropriate pipeline should be selected according to the experimental platform and library design.
- Barcode processing also requires distinguishing genuine cell-associated barcodes from background barcode sequences. Not every barcode observed in a sequencing library necessarily represents a cell. Some droplets may contain ambient RNA released from damaged cells, while others may contain very few RNA molecules. Computational methods such as EmptyDrops and the cell-calling procedures in Cell Ranger help identify barcodes likely to represent real cells. Researchers should also consider doublets, which occur when two or more cells are captured under a single barcode. A doublet can produce a mixed gene-expression profile and may be incorrectly interpreted as a distinct cell type. Tools such as Scrublet, DoubletFinder and scDblFinder can help identify potential doublets.
- Errors in barcode or UMI sequences can arise during sequencing, reverse transcription or library amplification. A sequencing error may cause a genuine barcode to appear as an invalid sequence or make one UMI appear different from its original form. Barcode correction uses the known structure of the barcode system and the set of valid barcode sequences to identify and correct likely errors. UMI correction commonly groups similar UMI sequences associated with the same cell and gene, using sequence similarity and read-count relationships to distinguish likely errors from independent molecules. However, correction must be applied carefully because two genuinely different molecules can occasionally have similar UMI sequences.
- UMI-based counts are often more suitable for comparing gene expression than raw read counts in protocols designed for molecular counting. Nevertheless, UMI counts are not perfect measurements of the true number of transcripts in a cell. RNA capture efficiency, reverse transcription efficiency, sequencing depth, transcript abundance, RNA degradation and sampling variation all affect the number of detected molecules. Some transcripts may not be captured at all, producing zero counts even when they are biologically present. This phenomenon contributes to the sparsity of single-cell gene-expression matrices and is commonly discussed in relation to dropout and detection sensitivity.
- Cell barcodes and UMIs also support advanced analyses beyond basic gene-expression quantification. They enable researchers to identify cell populations, compare gene-expression profiles, investigate cellular development and differentiation, detect rare cell types and integrate data from multiple samples. Sample-specific or multiplexing barcodes can be introduced to distinguish cells originating from different biological samples or experimental conditions. This makes it possible to pool samples during library preparation and reduce some technical variation, provided that sample identity is preserved and correctly assigned during analysis.
- In animal genomics and breeding research, cell barcodes and UMIs support the analysis of gene expression in individual cells from tissues associated with economically important traits. Researchers can investigate cell populations involved in muscle growth, milk production, reproduction, immunity, feed efficiency and disease resistance. For example, single-cell RNA sequencing can identify cell-specific expression patterns in mammary tissue during lactation or in muscle tissue during development. UMI-based quantification helps estimate the relative abundance of captured transcripts, while cell barcodes preserve the identity of the cells in which those transcripts were detected. Integration with genomic variation, quantitative trait loci, epigenomic information and phenotypic data can provide further insight into the molecular mechanisms underlying complex traits.
- Unique molecular identifiers and cell barcodes are therefore fundamental to modern single-cell RNA sequencing. Cell barcodes preserve the cellular origin of captured RNA molecules, while UMIs distinguish original molecules from their amplified copies. Together, they enable accurate molecular counting, cell-specific gene-expression analysis and the reconstruction of complex cellular populations. Their correct use, together with appropriate computational processing and quality control, is essential for reliable single-cell transcriptomic research.