Modern analytical instruments possess immense operational capacity. A single next-generation sequencing lane, high-density microplate, or solid-state nanopore array can generate millions of data points in a single run. However, running just one biological sample per run is economically and logistically impractical; the fixed overhead costs of reagents, labor, and instrument time create a severe bottleneck for large-scale studies.
To resolve this mismatch, researchers use sample multiplexing: combining multiple distinct biological samples into a single physical reaction vessel or instrument run, and then computationally sorting them afterward. This approach trades off individual per-sample depth or redundancy for massive gains in throughput and cost efficiency.
The Core Mechanism: Barcoding and Demultiplexing
How do we mix dozens or hundreds of samples without losing their individual identities? The primary mechanism is molecular barcoding. Before samples are pooled, a unique, identifiable sequence tag (a barcode or index) is physically attached to the molecules of each sample.
As Isabel A. S. Bonatelli et al. describe in Using Next Generation RAD Sequencing to Isolate Multispecies Microsatellites for Pilosocereus (Cactaceae), attaching unique DNA sequences to individual genomic libraries before pooling allows for the simultaneous processing of multiple individuals or species. Once the pooled library is sequenced, downstream bioinformatics algorithms read these tags and sort the data back to their original sources—a process known as demultiplexing.
The timing and method of barcode introduction can vary depending on the assay's workflow:
- Primer-Based Barcoding: Emma Collins et al. introduced an internal barcoding approach in A next generation targeted amplicon sequencing method to screen for insecticide resistance mutations in Aedes aegypti populations reveals a rdl mutation in mosquitoes from Cabo Verde, where short tags are placed directly onto PCR primers. This allows hundreds of individual samples to be pooled immediately after amplification. Similarly, Wenli Shen et al. incorporated barcodes directly into single-cell amplification processes in Developing a microfluidic‐based epicPCR reveals diverse potential hosts of the mcrA gene in marine cold seep to track host-gene links.
- Ligation-Based Barcoding: In structural biology, Cameron Myhrvold et al. showed in Barcode extension for analysis and reconstruction of structures how single-stranded DNA ligation can attach barcodes to individual structural strands, enabling parallel tracking of hundreds of components.
- Non-Genomic Barcoding: This concept extends beyond traditional sequencing. In biophysics, Sarah E. Sandler et al. demonstrated in Multiplexed Digital Characterization of Misfolded Protein Oligomers via Solid-State Nanopores that distinct DNA barcodes allow multiple protein oligomer samples to be analyzed simultaneously within a single solid-state nanopore measurement, drastically accelerating therapeutic screening.
The Statistical Trade-off: Depth vs. Breadth
Multiplexing introduces a fundamental architectural trade-off: a fixed instrument budget (such as the total number of reads per sequencing lane) must now be shared among all pooled samples. Consequently, the individual data depth per sample decreases. Why does this work? Because for many biological questions, statistical power comes from sample size (breadth) rather than exhaustive depth per sample.
Joshua D. Campbell et al. explicitly evaluated this trade-off in Assessment of microRNA differential expression and detection in multiplexed small RNA sequencing data. They found that while deeper sequencing uncovers rare transcripts, fold-change estimates for moderately to highly expressed microRNAs remain highly reproducible even at high multiplex levels. For differential expression studies, expanding the biological sample size yields far more statistical benefit than deep sequencing a few individual samples.
This principle underpins low-pass whole-genome sequencing. Zirui Dong et al. validated in Low-pass whole-genome sequencing in clinical cytogenetics: a validated approach that multiplexing dozens of samples into an ultra-low coverage (~0.25x) run drastically cuts costs while maintaining clinical sensitivity for copy-number variants.
Similarly, in clinical non-invasive prenatal testing, Taylor J. Jensen et al. demonstrated in High-Throughput Massively Parallel Sequencing for Fetal Aneuploidy Detection from Maternal Plasma that higher sample multiplexing reduces labor and costs. Sophisticated downstream bioinformatics extract robust statistical signals from fewer reads, preserving clinical sensitivity while maximizing throughput.
Advanced Pooling Geometries and Workflows
To push the boundaries of throughput, researchers have moved beyond simple linear pooling (one barcode per sample) to more complex combinatorial and workflow-level pooling strategies.
Combinatorial Pooling
Instead of assigning one unique barcode to one sample, Denisa Duma et al. developed a strategy in Combinatorial Pooling Enables Selective Sequencing of the Barley Gene Space where each DNA clone is distributed across a unique set of overlapping pools. By sequencing these pools and applying a deconvolution algorithm, the original source of each read can be mathematically reconstructed. This acts like a coordinate system, dramatically reducing the total number of sequencing reactions required to map complex genomes.
Workflow-Level Pooling
Pooling can also be performed early in the experimental pipeline to eliminate labor bottlenecks. Bruce James Wittmann et al. introduced a "pooling-first" workflow for protein variant libraries in evSeq: Cost-Effective Amplicon Sequencing of Every Variant in a Protein Library, where entire multi-well plates of PCR products are combined before purification and quantification. This relies on the intuition that variation in PCR yield across a plate is minimal enough to maintain data quality, bypassing tedious individual sample preparation.
Downstream Capture Pooling
Multiplexing can also optimize intermediate enrichment steps. Meiyan Li et al. showed in Performance comparison of four exome capture platforms on DNBSEQ-Series high throughput sequencer that pooling multiple DNA libraries into a single whole-exome capture reaction (such as an 8-plex configuration) maximizes the utility of expensive capture reagents and significantly reduces per-sample enrichment costs.
Through these diverse pooling geometries, large-scale comparative studies—such as the massive transcriptomic atlas of bacterial pathogens developed by Kemal Avican et al. in RNA atlas of human bacterial pathogens uncovers stress dynamics linked to infection—can scale to handle dozens of species and environmental conditions simultaneously.
Key Foundational Work
Efficient yeast ChIP-Seq using multiplex short-read DNA sequencing (2009) — contributed 1 concept elements to this direction.
An Improved Method for Evaluating Serum Antibody Titers. (1991) — contributed 1 concept elements to this direction.