COMPASS Technology
Delivering deeper patient classification through comprehensive molecular signal analysis
Magellan’s COMPASS platform combines data-rich serum or plasma with advanced mass spectrometry and purpose-built computational analysis.
Designed to capture and interpret the full molecular signal produced by modern instruments, it represents a more complete use of blood-based data—enabling a clearer understanding of clinical trial patients.
What COMPASS Unlocks
Magellan’s COMPASS platform removes the core limitations that have historically constrained mass spec–based discovery. By overcoming these barriers, it allows researchers to fully harness the power of modern mass spec instruments—at a scale and depth that was previously unattainable.
COMPASS lets mass spectrometry operate at its full potential.
Key Challenge: Data Overload
Modern mass spectrometers can detect over one million molecular species in a single complex biological sample. However, traditional software cannot process data at this scale—especially when comparing across dozens, hundreds, or thousands of samples.
As a result, the entire research workflow is constrained. Researchers are forced to:
-
Reduce sample numbers and study size
-
Modify sample preparation to reduce molecular complexity
-
Alter mass-spec run settings and methods to constrain output
-
Pre-filter or bias data before software analysis
These constraints limit discovery from the outset.
Mass spec generates massive data—but legacy software forces researchers to leave most of it behind.
Magellan’s Solution: COMPASS
Magellan developed COMPASS, a purpose-built machine learning platform designed to process and analyze the complete signal output of modern mass spectrometry.
Instead of forcing data to fit software limitations, COMPASS is built to match the true scale and complexity of the data—enabling full-spectrum analysis without compromise.
COMPASS turns mass spectrometry from a constrained tool into a complete system for biological discovery—capturing, analyzing, and interpreting data at its full depth.
Key Advantages Enabled by COMPASS
By removing computational constraints, COMPASS fundamentally changes how mass spec studies can be designed and executed:
-
Maximized sample preparation: Samples can be prepared to release the broadest possible range of molecular species
-
Unbiased data capture: Instruments can detect all measurable signals without pre-filtering or artificial constraints
-
Flexible study scale:
-
Deep insights from small cohorts (~20 samples)
-
Scalable to thousands of samples—while still analyzing full data for each
-
This represents a shift from constrained experiments to truly comprehensive molecular profiling.
COMPASS removes tradeoffs—researchers no longer have to choose between depth, scale, or accuracy.
Machine Learning Capabilities
Analyzing one million molecular datapoints per sample is only the starting point. The critical next step is knowing which of those datapoints genuinely matter to your clinical question — and which are noise.
COMPASS applies machine learning specifically designed for this challenge. Rather than requiring a hypothesis about which molecules to measure, it evaluates the entire dataset and identifies which features show non-random separation between your patient groups.
The result is a ranked list of molecular signals — ordered by confidence, validated against what random variation alone would produce, and directly connected to the clinical outcome you defined.
This is not pattern recognition for its own sake. Every signal COMPASS identifies is measured against a statistical threshold that distinguishes true biological separation from the natural variation present in any dataset. Signals that don't clear that bar are discarded. Those that do represent genuine molecular differences between your patients — differences that conventional approaches, working from a fraction of the same data, would never have seen.
COMPASS does not simply report what molecules are present. It identifies what is molecularly different between your patients — and provides a statistically validated confidence measure for each finding.
COMPASS finds patterns humans and traditional tools simply cannot see.
Results Delivered by the COMPASS System
The combination of full-spectrum data capture and advanced machine learning translates into meaningful clinical and research outcomes:
-
Clear patient stratification
-
Identification of treatment-response signals
-
Discovery of clinically relevant biomarkers
-
Direct answers to sponsor-specific research questions
COMPASS turns complex data into actionable biological and clinical insight.