FOUNDATION STUDIES

WHAT IS A FOUNDATION STUDY?

​

A Foundation Study is designed around your clinical questions: a comparison between or among groups, with each sample analyzed at a depth that makes meaningful results achievable despite the relatively small sample size.

Common comparisons include responders vs. non-responders, pre-treatment vs. post-response, early-stage vs. late-stage, or any two clinically defined patient populations. The clinical question is yours — COMPASS identifies the blood signature that separates them. 

No new sample collection is required. No changes to existing protocols. The Foundation Study begins with what you already have.

​

*Foundation Studies can also be conducted on samples distributed across a spectrum of values or outcomes, not only discrete two-group comparisons.

POWERED BY DEPTH, NOT SAMPLE COUNT

​

The statistical confidence of a COMPASS Foundation Study comes from data depth, not sample count alone. Analyzing an average of one million molecular datapoints per sample means a 20-sample Foundation Study generates twenty million molecular observations — a scale at which genuine signal consistently separates from random noise.

Conventional biomarker studies rely on large sample numbers to compensate for the limited number of analytes measured per sample. When a panel measures dozens or hundreds of features, larger sample sizes are needed to achieve statistical power. COMPASS inverts that relationship. The power comes from what is measured per sample — not from how many samples are in the study.

The machine learning identifies which features genuinely discriminate your groups across those twenty million observations — distinguishing real signal from noise with high confidence at this scale.

If discriminating blood signatures exist in your patient population, a Foundation Study is designed to find them. The findings are reliable and meaningful at this sample size.  A larger confirmatory study adds replication confidence, though the results themselves are usually not materially changed.  

THE PROCESS

​

Step 1 — Discovery Conversation (30 minutes). We learn about your program: the indication, trial design, patient population, and the clinical questions you need answered. We will tell you directly whether we think a Foundation Study is likely to be informative for your situation.

​

Step 2 — Sample Evaluation.  You share details about your banked or prospectively collected samples — volume, collection timepoints, storage conditions. We confirm suitability and define the analysis parameters together.

​

Step 3 — COMPASS Analysis.  Sample aliquots (minimum 100µL) are processed to enrich for low-abundance blood peptides and analyzed by LC-MS. COMPASS processes the complete mass spectrometry output — an average of one million molecular datapoints per sample — and applies machine learning to identify the features that distinguish your clinical groups.

​

Step 4 — Results.  We deliver a detailed report identifying the molecular features that separate your patient groups, ranked by machine learning confidence. Signal is clearly distinguished from noise. All raw and processed data is transferred to you. All discovered markers are your intellectual property.

​

Step 5 — Optional Next Steps.  Foundation Study results are meaningful and reliable as a standalone finding. However, sponsors may choose to pursue one or both of the following:

​

• Molecular identification of top-ranked markers through targeted MS/MS fragmentation analysis. This identifies the specific molecules behind the blood signatures and adds biological context to the clinical findings. Identification is attempted but cannot be guaranteed on all markers

• A confirmatory or extended study with higher sample numbers to build replication confidence 

• Additional Foundation Studies on different samples, asking different clinical questions 

WHAT YOU RECEIVE

​

• A molecular feature map distinguishing your patient groups, ranked by decision value

• Machine learning confidence assessment — clearly distinguishing signal from random noise

• Full data transfer — all raw and processed mass spectrometry data belongs to you

• IP ownership — any molecular features identified as clinically relevant are your intellectual property

• A written recommendation on logical next steps, whether further molecular characterization, additional study, or neither

INVESTMENT

​​​

Start with a Foundation Study — defined scope, defined cost, defined timeline.

​

For around $20,000, a Foundation Study delivers a reliable answer to your most pressing clinical question about your patient population. That answer is grounded in twenty million molecular observations — an average of one million datapoints per sample — which is why the findings are meaningful and reliable at this contained investment level.

​

​The Foundation Study stands on its own as a defined engagement with a clear deliverable — a reliable answer grounded in deep data, at a constrained and predictable cost. Sponsors may then elect to proceed with a Confirmatory Study to validate the findings at higher sample numbers.

IS A FOUNDATION STUDY RIGHT FOR YOUR PROGRAM?

​​

A Foundation Study tends to be most valuable when:

​

• A past or current trial showed a signal in a subgroup, but the patients driving that signal cannot be distinguished prospectively — and a blood-based marker to identify them going forward has not been established

• The program is approaching a transition between phases and the molecular basis for patient selection or stratification has not yet been characterized

• You have banked samples from a program that did not advance and want to understand whether a molecular characterization of the patient population could inform a repositioning or next step

• Prior targeted biomarker panels returned limited or no meaningful signal — and the question of whether a deeper signal exists remains open

• A clinical question about your patient population has not been answerable with the tools available so far

WHAT MAKES THIS DIFFERENT FROM A PROTEOMICS CRO?

​​

Most proteomics services run panels — predefined sets of known protein targets analyzed against your samples. That approach can only find what you already know to look for. If the blood signatures distinguishing your clinical groups have not been previously characterized, a panel will not reveal them.

Magellan starts from the complete mass spectrometry output — an average of one million molecular features per sample — and asks the data what is different between your clinical groups. No predefined targets. No pre-filtering. If a signal exists in blood at the depth COMPASS analyzes, it will be found.

Even if you have run mass spectrometry studies before, this is different. The scale of data retained, the absence of pre-selection, and the machine learning approach to identifying signal within that scale are what change the outcome.