NGS-Based Proteomics Platforms Compared
Turning proteins into barcodes: a guide to NGS-based proteomics
Welcome back to Field Signals, my series on what I’m seeing and hearing out in the field. The first four instalments, written after attending the Festival of Genomics in Boston and EACR in Budapest, touched on where genomics is heading, the convergence of multiomics, AI integration and a shift in experimental models, and why mechanistic biology still matters. One previously discussed theme was proteomics expanding alongside genomics, so this time I wanted to take a closer look.
I recently attended the Human Proteome Organization meeting (HUPO 2026), and as someone with a genomics background, I found it a fantastic way to get better acquainted with the proteomics field. Mass spectrometry remains the gold-standard method for unbiased proteomics discovery, but what I found particularly interesting was the prevalence of next-generation sequencing (NGS)-based proteomics. Coming from my previous experience at 10x Genomics, I see proteomics as a natural next step to complement transcriptomics.
One of the historical limitations of proteomics is that, unlike nucleic acids, proteins can’t be amplified. NGS-based proteomics offers a way around this limitation. In this article, I will cover four NGS-based proteomics technologies (Olink, SomaSeq, NULISA, and ProteinXI), explain how they differ, and offer a framework for matching them to your research. At their core, all four are affinity-based platforms that convert protein detection into DNA barcodes, then count those barcodes using NGS. Because many researchers already have access to the sequencing infrastructure these platforms rely on, they offer an accessible entry point into high-plex protein profiling.
Four NGS-based proteomics platforms and how they work
The four platforms I will compare are Olink (Thermo Fisher), SOMAmer Proteomics (Illumina), NULISA (Alamar Biosciences), and ProteinXI (Meso Scale Discovery):
- Olink:
Proximity Extension Assay. Each target protein is recognised by a matched pair of DNA-tagged antibodies. When the matched antibody pair binds the same protein, their DNA tags are brought close enough to hybridise. The hybridised DNA is extended to create a barcode unique to that protein. These barcodes are amplified and counted by sequencing, with read counts reflecting relative protein abundance. - SOMAmer Proteomics:
This is the aptamer chemistry behind SomaScan, now available as SomaSeq through Illumina. Target proteins are recognised by SOMAmers (slow off-rate modified aptamers), single-stranded DNA molecules with modified nucleotides that fold to specifically bind a target. SOMAmers attached to beads capture their target proteins. Because SOMAmers release their targets slowly, a competitor added during incubation prevents loosely bound, non-specific proteins from rebinding, so only tightly bound complexes are carried through. The SOMAmers are then released and become the barcode. They are indexed, amplified, and counted by sequencing, with read counts reflecting relative protein abundance. - NULISA:
Each target protein is bound by a matched pair of DNA-tagged antibodies to form a complex. The complexes are then purified twice: first captured on beads through the capture antibody’s DNA tag, washed and released, then recaptured on a second set of beads through the detection antibody and washed again. This sequential capture and release removes unbound antibodies and non-specific background. The paired DNA tags are then joined into a single reporter that acts as a barcode unique to that protein. The barcodes are amplified and counted by sequencing, with read counts reflecting relative protein abundance. Calibration curves can also be run on selected panels to convert read counts for some targets into absolute concentrations. - ProteinXI:
Each target protein is recognised by a matched pair of antibodies. The capture antibody sits on a bead, which also carries DNA tags with “lock” molecules. Detection antibodies carrying DNA tags with “key” molecules are then added to bind the captured proteins. The key only fits the lock when the capture and detection antibodies on the same protein are a matched pair. Matched lock and key complexes are then released from the beads, and the combined DNA tags from both antibodies form the barcode, which is counted by sequencing. Calibration curves are also run in the assay, so read counts are converted into absolute concentrations across the whole panel.
Olink vs SomaSeq vs NULISA vs ProteinXI: comparison table
In the table below, I compare the largest panels currently offered on each platform, using publicly available information from vendor websites and supporting resources.
| Olink (Thermo Fisher) | SOMAmer Proteomics (Illumina) | NULISA (Alamar Biosciences) | ProteinXI (Meso Scale Discovery) | |
|---|---|---|---|---|
| Detection method | Matched pair of DNA-tagged antibodies (Proximity Extension Assay) | SOMAmers: single-stranded DNA aptamers with modified nucleotides | Matched pair of DNA-tagged antibodies | Matched pair of antibodies with DNA tags carrying a lock (capture) and key (detection) |
| Background removal | Dual recognition: DNA tags only hybridise and extend when both antibodies bind the same protein | Slow off-rate binding plus a competitor during incubation that prevents non-specific rebinding | Sequential capture and release on two sets of beads with washes to remove unbound antibodies and background | Key only fits lock for a matched antibody pair; only matching complexes are released from beads |
| Barcode | Hybridised and extended DNA from both antibody tags | The SOMAmer | DNA tags from both antibodies joined into a single barcode | Combined DNA tags from both antibodies |
| Largest panel | Explore HT: 5,400+ proteins | SomaSeq Discovery: ~9,500 proteins | Immune 340 | Human Discovery 400 (600 proteins at launch*) |
| Minimum sample input† | 2 µL | 55 µL (35 µL for cell lysate & tissue homogenate) | 25 µL | 10 µL |
| Quantitation | Relative | Relative | Relative | Absolute, using calibration curves |
| Recommended sample types | Plasma, serum, CSF | Plasma, serum (cell lysate & tissue homogenate with additional kit) | Plasma, serum | Plasma (human biofluids) |
| Claimed dynamic range‡ | fg – mg/mL (10 logs) | fM – µM (>10 logs) | aM – µM (12 logs) | fg – mg/mL (12 logs) |
| Maximum sample throughput per run | Up to 344 samples | Up to 340 samples (if using NovaSeq X Plus) | Up to 86 samples | Not yet published |
| Additional instrumentation required | Liquid handling, sequencer | Liquid handling, Illumina sequencer | ARGO HT system, sequencer | Not yet published; requires sequencer |
| Commercial status | Established | Established | Established | Early access |
Information accurate as of October 2026.
* Information from Meso Scale Discovery’s HUPO 2026 session.
† Vendor-stated volumes; service providers typically request additional volume (e.g. 40 µL for Olink Explore HT).
‡ As stated on vendor websites.
How to choose an NGS-based proteomics platform
Before getting into the relevant questions, it’s worth stating that these platforms are not a replacement for mass spectrometry or other proteomics technologies. Mass spectrometry remains the established method for truly unbiased discovery, including proteins outside any predefined panel, post-translational modifications, proteoforms, and non-model organisms. The NGS-based platforms highlighted in this article are affinity-based and measure a defined set of targets. The best choice is often more than one technology, used for what each does best. With that in mind, here are the questions I would consider.
Breadth or depth?
The first question is whether you need a broad survey of the proteome or deep coverage of a specific biological area. SomaSeq Discovery and Olink Explore HT are built for breadth, covering thousands of proteins in a single run. NULISA takes the opposite approach, with curated panels focused on areas such as immunology and neurology. ProteinXI sits in between, with a broad discovery panel of several hundred proteins. The ability to measure more targets does not automatically make a technology better. A focused panel can mean fewer multiple-testing penalties and more sensitive measurements for the targets you actually care about.
Relative or absolute quantitation?
Most of the NGS platforms I’ve covered report relative abundance, which is well suited to comparing groups within a study. However, this kind of measurement becomes limiting when you want to compare across studies, combine datasets run at different times, or relate your results to clinically meaningful thresholds. Absolute quantitation, reported as concentrations, makes those comparisons more straightforward. ProteinXI reports absolute concentrations across its panel using calibration curves, and Alamar offers absolute quantitation for a subset of targets on one of its panels. For the other platforms, absolute values for key hits will typically require follow-up with orthogonal methods.
How low do you need to go?
Many of the most biologically interesting proteins, such as cytokines and neurological markers, circulate at very low concentrations. If your hypothesis depends on these, sensitivity matters more than plex. NULISA has made sensitivity its core claim, and each vendor publishes claims for its most sensitive assays. Rather than comparing headline numbers, it’s important to check whether the specific proteins you care about are reliably detected in your sample type.
Antibodies or aptamers?
Olink, NULISA, and ProteinXI all rely on matched antibody pairs, so a signal requires two independent binding events on the same protein, which helps limit off-target signals. SOMAmer Proteomics uses a single aptamer per target, which is part of what allows such broad coverage. The trade-off is that some measurements don’t always agree closely with antibody-based assays for the same protein. Neither approach is exempt from issues such as genetic variants affecting binding, which is why validating key findings with an orthogonal method is always good practice.
Sample volume and matrix.
Required sample volumes range from a few microlitres to over 50 µL, which can be decisive for biobank aliquots, paediatric samples, or CSF. Just as important is the list of validated sample types. Plasma and serum are well supported across these platforms, but if you work with CSF, tissue lysates, or other matrices, it’s essential to check what each vendor has validated before committing precious samples.
In-house or service?
All four platforms require access to a sequencer, and most also require dedicated automation or liquid handling to run at scale. For many groups, a core facility or service provider is the most practical starting point, particularly for a first study. Running these platforms in-house makes more sense when you have regular, high sample volumes and the infrastructure to support them.
What happens after discovery?
Discovery is rarely the end point. Before you start, it’s worth considering how you’ll validate your hits. Olink offers smaller targeted panels, and established targeted platforms such as ELISA, ECL, Simoa, and bead-based assays are widely used for validation. ProteinXI has shown strong concordance with Meso Scale Discovery’s ECL immunoassays, which provides a direct route from discovery to validation. Planning this step early in the experimental design process can save considerable time and effort later.
What drives cost?
Rather than quoting prices, which vary by region, volume, and service model, it’s more useful to understand what drives them. Plex affects reagent cost and the sequencing depth needed per sample. Sample numbers and batch sizes matter, since partially filled runs are less efficient. Running in-house brings instrument and staff costs, while service providers bundle these into a per-sample price. Controls, repeats, and replicates will also contribute to costs.
Final thoughts on NGS-based proteomics
There’s no single best NGS-based proteomics platform. Each technology has its strengths and weaknesses, and the right choice depends on your research question, your samples, and how you plan to take your findings forward. Often the strongest study design combines more than one approach, each addressing a different aspect of the research question.
What struck me most at HUPO was how quickly the field is broadening. Sequencing is being used to read out proteins not just in bulk biofluids, but also at single-cell resolution through approaches such as CITE-seq. In tissue, there is a growing range of imaging-based platforms that map proteins in their spatial context, sometimes alongside RNA. Meanwhile, single-molecule technologies from companies such as Nautilus, Quantum-Si, and Portal Biotech are opening up different routes into the proteome. These are beyond the scope of this article, but are well worth watching.
For anyone coming from a genomics background like mine, it’s an exciting time. The tools and infrastructure many of us already use every day are becoming a gateway to the proteome, and to the proteins that ultimately drive biology.