What a SomaScan Costs, and What You Get for the Money

A SomaScan run costs roughly $800–900 per sample at list price for the 11K menu when you go through an academic core, plus a batch minimum. The University of Houston core is one of the few sites that publishes numbers. It lists $850 per sample for academic clients and $875 for industry on the 11K platform, with data analysis as a $75 add-on. It quotes 7K assays in blocks of 85 samples. SomaLogic itself does not publish per-sample pricing for its fee-for-service lab. You get a quote, and the quote depends on the following.

• plex (Select 4K, 7K, 11K)
• sample count
• matrix
• whether you want their analysis deliverables

For a single individual buying one plasma sample of 11K proteomics, expect all-in cost closer to $1,200–1,600 once you add shipping on dry ice, blood draw and plasma processing, and the fact that you are buying one well on a plate someone else is filling.

That is the number. The more useful question is what drives it, and whether 11,000 aptamer measurements on one timepoint is the right thing to buy.

Why the per-sample number is not the cost

Three things inflate the real cost above the sticker price.

Plate economics. The assay runs in 96-well format with calibrator, buffer, and QC wells occupying part of every plate, which is why cores quote in units of ~85 study samples. If you have one sample, you are a passenger on a plate configured for someone else’s study, and you inherit their plate as your batch reference. If you have 20 samples across two years of longitudinal sampling, the right move is to freeze everything at −80 °C and run all of it on one plate at the end. Batch effect between plates is larger than the biological drift you are usually trying to detect.

Preanalytics. SomaScan wants EDTA plasma, on the order of 55–150 µL depending on menu and dead volume, never thawed more than necessary. Time-to-spin, tube type, and freeze-thaw count move hundreds of analytes, especially platelet and coagulation proteins. A pilot of at-home collection kits for proteomic studies found that storage and shipping conditions measurably shift the measured proteome, which is the main argument for a controlled draw and immediate processing when you can get one.¹ If you are paying $850 for the assay, do not save $40 on the collection.

Analysis. The $75 per-sample analysis add-on gets you a normalized matrix and standard QC plots. It does not get you anything you can act on. Budget your own time instead.

How the assay works, and why that shapes the price

SomaScan uses SOMAmers, chemically modified DNA aptamers with protein-like side chains on the nucleobases that give them slow off-rates against protein epitopes. Plasma is incubated with the full reagent pool on streptavidin beads, and unbound protein is washed away. A polyanionic competitor removes low-off-rate complexes. The SOMAmers are photocleaved off the bead and captured a second time by the protein they are bound to. They are then eluted and quantified by hybridization to an array or by sequencing readout. The output is relative fluorescence units per SOMAmer. It is not a concentration.²

The economics follow from that design: one reagent pool, one workflow, and plex is nearly free to expand. Going from 7K to 11K costs the vendor almost nothing in labor, which is why the price difference between menus is small compared to the difference in coverage. It also explains the failure mode. Every target is assayed under identical buffer and wash conditions, so any single SOMAmer may be measuring off-target binding, and you cannot tell from the RFU value alone.

That matters for interpretation. Head-to-head comparisons of the aptamer platform against antibody-based Olink used genetics and clinical traits as arbiters. They found that agreement for overlapping targets is modest overall and strongly bimodal. A subset of proteins correlate tightly across platforms, and a large subset barely correlate at all.³ The ARIC study reached the same conclusion on a large population sample.⁴ In CSF, comparison of SOMAscan against validated immunoassays for Alzheimer’s biomarkers showed good concordance for some analytes and poor concordance for others.⁵ The practical rule: a SomaScan value is trustworthy in proportion to external evidence for that specific reagent, chiefly a cis-pQTL in the gene encoding the protein.

Getting more out of the money

Two cheap moves increase the value of a single run.

Use your genome as a validity filter. Large pQTL atlases now map genetic variants to thousands of plasma proteins, including a 7,000-protein atlas in East Asian cohorts.⁶ If a SOMAmer has a strong cis-pQTL and your own genotype at that variant predicts your measured level in the expected direction, the reagent is probably measuring the intended protein in you. If a reagent has no cis signal anywhere in the literature, treat a single extreme value as a lead. Do not treat it as a measurement. Published genetic score atlases for multi-omic traits let you go further and ask whether your measured level departs from what your genotype predicts, which is where the interesting signal lives.⁷

Use replicates in time rather than plex. Almost every interesting inference from proteomics is about change. Cross-platform work on extreme old age found that the signatures that replicate across platforms are the ones driven by large, coherent shifts across related proteins. Signatures driven by single-analyte outliers do not replicate.⁸ Trauma studies with serial sampling show the same thing: the readable signal is the trajectory of chemokine and cytokine modules over hours and days. It is not one RFU value.⁹ Two timepoints of the Select 4K menu often teaches you more than one timepoint of 11K.

SomaScan versus Olink on price per protein

Olink Reveal measures about 1,000 proteins with an NGS readout and was explicitly positioned as the low-cost entry into their portfolio. Olink does not publish list pricing either. Cores generally quote Reveal well under half the per-sample cost of SomaScan 11K, in the low hundreds of dollars at volume. Olink Explore HT sits at roughly 5,400 proteins and prices closer to SomaScan.

Per protein, SomaScan 11K is the cheapest thing on the market. Per trustworthy protein, the gap narrows a lot, because proximity extension assay uses two antibodies per target and gets specificity from requiring both to bind. If you want breadth and hypothesis generation, buy SomaScan. If you want a smaller number of measurements you will act on with a clinician, a targeted immunoassay panel or a clinical lab test is the better spend. SomaScan is research use only. Nothing in an ADAT file is a diagnosis. If something in it concerns you, the next step is a validated clinical assay ordered by a physician. The next step is not a second aptamer run.

What comes back, and how to work with it

The deliverable is an .adat file: a flat text file with a header block of run metadata, a column-metadata block keyed by SeqId (seq.10000.28 style identifiers, with UniProt, gene symbol, target name, and dilution group), row metadata per sample, and then the RFU matrix. Read it with the vendor’s R package:

library(SomaDataIO)
adat <- read_adat("SS-2500xx.hybNorm.medNormInt.plateScale.calibrate.anmlQC.qcCheck.adat")
ann  <- getAnalyteInfo(adat)          # SeqId -> UniProt, gene, dilution, ColCheck
x    <- log2(adat[, getAnalytes(adat)])

Read the filename. Each suffix is a normalization step that was applied: hybNorm (hybridization control normalization), medNormInt (intraplate median signal normalization), plateScale, calibrate, and anml (adaptive normalization by maximum likelihood against a reference population). ANML is the one to think about. It pulls your sample toward a healthy reference distribution, which suppresses plate effects and can also suppress real global shifts. For a single-individual longitudinal profile we keep both the ANML and the median-normalized versions and check that any finding survives in both.

QC fields that matter: RowCheck and per-sample NormScale_* factors in row metadata, and ColCheck per analyte. SomaLogic’s guidance is to scrutinize samples whose normalization scale factors fall outside roughly 0.4–2.5, and analytes flagged at the column level should not be interpreted at all. Work in log2 space, quantify dilution-group membership (40%, 1%, 0.005% plasma) when comparing across analytes, and expect intra-assay CVs in the 5–15% range for well-behaved reagents and much worse for the rest.

For an applied example of what a well-designed SomaScan study yields, the ACC-inhibitor NASH work used the platform to find pharmacodynamic markers and then followed the mechanism with independent measurement.¹⁰ That pattern, broad discovery followed by targeted confirmation, is what the $850 is for.

Footnotes

1. Caroline Scranton, Xiaoxiao Sun, Dominic Rodriguez, et al. Pilot Study to Determine the Efficacy, Feasibility, and Impact of Storage Conditions on At- Home Blood Collection Kits for Proteomic Studies. 2026. https://doi.org/10.21203/rs.3.rs-8737461/v1 ↩

2. Hanjun Zhao, Laura Winchester. High Throughput Proteomics Using the SomaLogic Platform. Methods in Molecular Biology, 2025. https://doi.org/10.1007/978-1-0716-4462-1_13 ↩

3. Daniel H. Katz, Jeremy M. Robbins, Shuliang Deng, et al. Proteomic profiling platforms head to head: Leveraging genetics and clinical traits to compare aptamer- and antibody-based methods. Science Advances, 2022. https://doi.org/10.1126/sciadv.abm5164 ↩

4. Mary R Rooney, Jingsha Chen, Christie M Ballantyne, et al. Comparison of Proteomic Measurements Across Platforms in the Atherosclerosis Risk in Communities (ARIC) Study. Clinical Chemistry, 2022. https://doi.org/10.1093/clinchem/hvac186 ↩

5. Jigyasha Timsina, Duber Gomez-Fonseca, Lihua Wang, et al. Comparative Analysis of Alzheimer’s Disease Cerebrospinal Fluid Biomarkers Measurement by Multiplex SOMAscan Platform and Immunoassay-Based Approach. Journal of Alzheimer’s Disease, 2022. https://doi.org/10.3233/jad-220399 ↩

6. Alfred Pozarickij, Baihan Wang, Ahmed Mohamed, et al. Genomic atlas of 7,000 plasma proteins and their associations with diseases and traits in East Asian populations. 2026. https://doi.org/10.64898/2026.02.05.26345625 ↩

7. Yu Xu, Scott C. Ritchie, Yujian Liang, et al. An atlas of genetic scores to predict multi-omic traits. 2022. https://doi.org/10.1101/2022.04.17.488593 ↩

8. Eric R. Reed, Kevin B. Chandler, Prisma Lopez, et al. Cross-platform proteomics signatures of extreme old age. GeroScience, 2024. https://doi.org/10.1007/s11357-024-01286-x ↩

9. Jillian Bonaroti, Isabel Billiar, Hamed Moheimani, et al. Plasma proteomics reveals early, broad release of chemokine, cytokine, TNF, and interferon mediators following trauma with delayed increases in a subset of chemokines and cytokines in patients that remain critically ill. Frontiers in Immunology, 2022. https://doi.org/10.3389/fimmu.2022.1038086 ↩

10. Pitchumani Sivakumar, Michelle Saul, Douglas Robinson, et al. SomaLogic proteomics reveals new biomarkers and provides mechanistic, clinical insights into Acetyl coA Carboxylase (ACC) inhibition in Non-alcoholic Steatohepatitis (NASH). Scientific Reports, 2024. https://doi.org/10.1038/s41598-024-67843-8 ↩