OUR DISCOVERY TECHNOLOGY

Logo background removed.png

COMPASS

COMPASS is composed of a mass spectrometry protocol and AI software.  It has achieved: 

  • Identification of biomarkers that predict onset of Preeclampsia and Preterm Birth, and which diagnose or stage Alzheimer's Disease, Endometriosis, and Breast Cancer metastasis 

THE HIDDEN PROTEOME: Our unique data source

The contents of the blood are a window into the health of the organism

The story is mainly told in a small fraction of the serum: the “Hidden Proteome”

Our COMPASS technolgy has been developed to explore this fraction of the serum

Why the Hidden Proteome can deliver high quality markers:

  • Contains a large number of unique species: >120,000 peptides, small molecules, and lipids, which are "hidden" by high abundance proteins

  • Unlike DNA, the contents of the Hidden Proteome dynamically change and reflect the current health of the individual

COMPASS Process

1.

DRAW BLOOD, PROCESS TO SERUM

Simple blood draw and conversion to serum

2.

DISCARD ABUNDANT PROTEINS 

There are approximately 30 super abundant molecules that occur more that 10^7 more frequently in serum than other molecules. These are removed according to Magellan's licensed protocol

3.

SUBJECT SAMPLE TO MASS SPECTROMETRY

Mass spectrometry identifies molecular mass(m/z) and retention time(rt) for the molecules in the remaining serum. The result is a count of how abundant each such molecule is. These abundances form the base feature set for our analysis

4.

COMBINE MASS SPEC DATA WITH PATIENT DATA

The mass-spec data is then combined with data known about the patient. This includes demographics, co-morbidity and any relevant test results. Most important is the outcome we are trying to predict.

5.

APPLY MACHINE LEARNING TO DEVELOP A PREDICTIVE MARKER SET

Our proprietary machine learning is then applied to the combined data set to train a function that can predict the desired outcome. The resulting trained classifier also contains information about which data contribute to the decision.

Discovery Process

In many cases, the following approach is followed for discovering predictive markers:

 

  1. Study the biology of the condition

  2. Identify one or more molecules that are critical to the process

  3. Develop an assay for detecting the identified molecules

  4. Use the assay to try to predict the desired condition

  5. If this did not work, restart at step 1

The above process requires significant biological skill, capabilities, time, and expense

 

The Magellan approach works differently:

 

  1. Identify a set of patient the exhibit the condition or may exhibit the condition in the future, as well as a set of control patients

  2. Extract blood, produce serum and perform Magellan's mass-spec analysis on all patients

  3. Using machine learning, analyze >120,000 unique species, and develop a classifier for the desired outcome

In some cases, the identity of the markers can be elucidated via MS2, or further biochemical analysis.   

 

The Magellan method does not require an underlying understanding of the biology, is target agnostic, and is faster, less expensive than traditional approaches.

© 2018 by Magellan Bioanalytics