Unique Molecular Identifiers (UMIs)

A unique molecular identifier (UMI) — also called a molecular barcode or molecular index — is a short, random or semi-random nucleotide tag, typically 6 to 12 base pairs long, ligated or appended to individual DNA or RNA fragments during library preparation, before PCR amplification. Because the tag is applied to the original template molecule rather than to a PCR product, every read generated later that carries the same UMI (usually combined with the fragment’s alignment coordinates) can be traced back to a single starting molecule. Bioinformatics tools then collapse all reads sharing a UMI into one consensus read, rather than counting each PCR copy as if it were independent biological evidence.

What UMIs Are Used For

UMIs solve problems that alignment-position-based duplicate marking (e.g. Picard MarkDuplicates) cannot solve on its own:

• True PCR/optical duplicate removal. In targeted panels and amplicon sequencing (see also PCR protocol basics), many independent template molecules legitimately start and end at the same genomic coordinates. Position-only deduplication would incorrectly discard some of these as duplicates, or fail to flag true PCR duplicates that happen to align differently. UMIs distinguish the two cases directly.
• Error correction. Building a consensus sequence across every read that shares a UMI averages out PCR-introduced and sequencer-introduced base-call errors, which is essential when the biological signal of interest is a rare variant present in only a small fraction of the original molecules.

Rare-variant and liquid-biopsy sensitivity

UMI-based error correction is what allows assays such as circulating tumor DNA (ctDNA) liquid biopsy panels and minimal residual disease (MRD) monitoring assays to reliably call variants at allele frequencies well below 1% — frequencies that are otherwise indistinguishable from routine PCR and sequencing error rates.

Absolute molecule counting

In single-cell and low-input RNA sequencing (for example, droplet-based 3′ gene expression workflows), each cDNA molecule receives a UMI in addition to a cell barcode. Because PCR amplification of a given cDNA molecule produces many reads that all share its UMI, the number of distinct UMIs observed per gene per cell — not the raw read count — approximates the true original transcript count. This corrects for PCR amplification bias, which otherwise skews apparent expression levels toward whichever transcripts amplified most efficiently.

The UMI approach was first formally described by Kivioja and colleagues in a 2011 Nature Methods paper on counting absolute numbers of molecules using unique molecular identifiers, and has since become standard practice across targeted sequencing, single-cell genomics, and clinical oncology NGS assays.

Procurement and Kit-Selection Considerations

Not every NGS library preparation kit incorporates UMIs, and UMI-enabled chemistries generally carry a cost premium over standard kits — the additional indexed adapter design adds manufacturing complexity, and the workflow requires UMI-aware downstream bioinformatics rather than a generic aligner-plus-deduplication pipeline. For a lab manager or procurement officer evaluating library prep kits (and tracking the resulting sample/reagent inventory — see barcode and RFID labeling for lab sample tracking), UMI capability is a specification worth comparing deliberately rather than assuming as a default. Relevant evaluation criteria include:

• UMI length and diversity. A longer, more diverse UMI sequence space reduces the risk of two unrelated molecules being assigned the same UMI (a "UMI collision") when multiplexing many samples or sequencing to high depth.
• Single vs. dual UMIs. Dual UMIs, applied to both ends of a fragment, generally provide stronger duplicate detection and error correction than a single UMI on one end, which matters more as the variant-detection threshold gets lower.
• Bioinformatics compatibility. Confirm the kit’s UMI structure is supported by an available, actively maintained analysis pipeline — whether an open-source tool such as UMI-tools or fgbio, or a vendor-supplied pipeline — and that the pipeline is compatible with your sequencing platform’s output format.
• Validation documentation for clinical or regulated use. If the assay will be used as (or as part of) a clinical laboratory test, the laboratory is responsible for its own CLIA analytical validation (limit of detection, reproducibility, and accuracy) specific to that kit’s UMI chemistry and your sample types — generic vendor marketing claims about sensitivity are not a substitute for assay-specific validation data, and any FDA clearance or authorization applies to the cleared kit and workflow as validated, not automatically to modified protocols.
• Lot-to-lot consistency. Because UMI incorporation efficiency and diversity affect downstream sensitivity, ask vendors for lot-specific quality-control data or a Certificate of Analysis, particularly for kits used in quantitative or low-input applications.
• Total cost at required sequencing depth. Rare-variant sensitivity from UMI-based consensus calling depends on having enough raw reads per UMI family (family size), not just enough UMIs. Budget for the sequencing depth the kit’s documentation specifies as necessary to reach your target limit of detection, not only the list price of the kit itself.

These are the same categories of evidence — documented validation data, lot traceability, platform compatibility, and total cost of ownership — that a genomics core facility or procurement office should request from any consumables vendor as part of standard due diligence, rather than relying on unverified comparative claims between named suppliers.