Haystack: Proteome molecular profiling for discovery and diagnosis

by John Wilson, PhD

Our technology reveals states, like cancerous or not, or drug responder or not. We do this by detecting many (10s of 1000s) of proteins and their variations and applying advanced analytics.
Huntington, NY, United States Diagnostics BigData Cancer

All Team Company Patients Physicians Hospital Partners Mission Innovation Details Supporters Comments Updates

About our project

The problem we solve:
Cancers are often found too late. Our technology allows screening for many kinds of cancer from blood with one test rapidly and economically. It can be extended to test for other states such as other diseases or drug response.
About our solution:
We analyze many, many proteins--where drugs work--with advanced statistics to find predictive and diagnostic signal where others cannot. Too often there is not one biomarker that indicates disease. We don't assume there is one and begin with serum, a familiar and easily accessed sample. We then process the proteins into peptides, or pieces of protein, with unique technology that provides speed and reproducibility (these have been bottlenecks). Next we analyze those peptides by mass spectrometry (LCMS-based proteomics), resulting quantitative data on 10s of thousands of protein features. We analyze these data with the most advanced multidimensional analytical platform ever created to generate a _panel_ of biomarkers that allows distinction of, say, cancerous and healthy, in the absence a single biomarker. This was done by Jeff for colorectal cancer--now a commercialized test--using 11 biomarkers. We are extending more advanced techniques to other cancers.
Progress to date:

Our technology is ready; we only need money to process samples. We have access to pancreatic and lung cancer samples sufficient to make blood tests for those cancers.

There are three components to our solution: 1) sample prep, 2) analysis on mass spectrometers and 3) data analysis. Proteomics has traditionally been fraught with huge variation in sample prep, which technology from the CSHL spinout ProtiFi has solved. [That technology, S-Trap sample processing, is now being used by the clinical proteomics field.] Mass spectrometers have become very good and we have enhanced their abilities though custom control and monitoring software; we have multiple labs which can run samples. Finally, our data analysis pipeline is ready.

Our collaborators at NYU and MSKCC are very enthusiastic about our approach and have simply asked "who's going to pay for the mass spec time." In sum, we have everything ready (including precious samples) and need money to generate data either in house or through third-party vendors.

We add that this approach has been proven and is on the market for colorectal cancer. That test, the SimpliPro Colon blood test, was developed by team member Jeff Jones. Making these tests requires extraordinary ability in statistical analysis.


About Our Team


Creator: John Wilson , PhD
Location: New York
Education: The Rockefeller University
Bio: I am a problem solver who learned many tools of science to be able to generate those solutions. To solve problems, I've started and lead new teams/organizations/companies (NYC Bio or ProtiFi, our CSHL biotech spinout, e.g.); invented or made: new proteins (Tryp-N); new devices; new software; new chemistries; new systems; etc. Biotech in particular is a powerful way to solve some really huge issues such as diagnosis and treatment of disease.
Hospital Affiliation: Cold Spring Harbor Laboratory
Title: CEO

About Team Members

Jeff Jones
CSO, PhD
Biography: Jeff developed the world’s only blood test for colorectal cancer, the SimpliPro Colon test. He Focuses on proteomics, statistics and systems biology to develop methods and characterize biological and clinical samples - in many cases he developed new analytical techniques. He is now applying edge machine learning to signal modeling and detection, developing algorithms to classify signal from noise in complex multi-omics data sets. He also dabble in predictive models using monte-carlo simulations.
LinkedIn: https://www.linkedin.com/in/jeff-jones-informatics

Darryl Pappin
VP, PhD
Biography: Darryl Pappin is a Professor at Cold Spring Harbor Laboratory where his lab develops chemical and computational methods for analysis of proteins and peptides typically via mass spectrometry. Darryl has developed search engines for mass spectrometry data that enable investigators to sift hundreds of thousands of experimental spectra at a time for database matches. He also seeks to reduce sample complexity via an approach termed chemical sorting to segregate classes molecule for specific analysis.

About Our Company

Haystack Inc.
Location: Huntington, New York
Product Stage: Prototype/MVP
YTD Sales: Less than $250,000
Employees: 3-5

How We Help Patients

The biggest issue in cancer is early detection: if identified early enough, you can beat it. The problem has been that many cancers (and especially highly deadly ones such as pancreatic cancer) are detected too late. Our technology at its heart provides this early detection and will save lives. In essence, our approach is a screen distilling multivariate molecular-level detail into answers like "This person is highly likely to have cancer even though they aren't currently presenting. They need immediate further medical analysis." Once the statistical analysis is understood, we can screen for more than one cancer from the same sample. We anticipate it will save both lives and money.

How We Help Physicians

Our technology assists providers first and foremost by allowing them to provide better care through earlier detection (especially of diseases like cancer) and, when applied to drug responder/non-responder status, by assisting them to give patients efficacious medications that will not cause adverse reactions. As, or even more importantly, our technology distills extraordinarily rich molecular-level details into meaning that aids physicians in care, in selection and use of both diagnostics and drugs. Our goal is to translate these data into better patient care.


How We Help Hospitals

Our technology benefits hospitals by increasing their success rate at detection of cancer. More people will come to hospitals using our technology. Our technology further assists by providing useful, understandable information to drive decisions in treatment and testing. We are a perfect partner for the pathology and diagnostic divisions of hospitals.


How We Help Partners

We are currently being approached by multiple big pharma companies as our technology can distinguis drug respondres from nonresponders which significantly increases the chances of FDA drug approval; these are perfect partners.

Overall, drugs with a companion diagnostic have a 3x better chance of FDA approval. However, like for cancer, there is typically no one biomarker that distinguises response (including adverse event) from positive outcomes. Our proprietary approach and especially our data analysis allows that distinction. Our approach also allows salvage of a failed drug trial by identification and segregation of the responding population.


Challenge Mission

Market Size

For cancer detection assays, the market sizes will be a function of the frequency of that cancer. Likewise, for drug responder assays, the market size will track closely with the rate of prescription. However, after we have developed multiple tests for multiple cancers, as one sample can then be used to screen for many cancers, we anticipate very wide adoption.

Our initial goal is to have the capacity to serve 20,000 proteomic profiles per year - achievable with a single analytical LCMS and workflow. That throughput has the ability to serve for approximately 70 pilot discovery projects balanced at 50:50 (150 samples per cohort) for a binary state predictor. Alternatively, approximately 12 medium sized clinical trials could be supported.



Projected 3 Year Growth

We plan to triple our total capacity every year (1, 3, 9), such that by year three we will have the approximate annual throughput of 180,000 samples. Accordingly, by end of year 3 we anticipate having sampled and analyzed nearly 250,000 proteomes, all mineable through our Knowledgebase. Our ultimate goal of sampling a million individuals will be within sight by end of year three.



How We Will Make Money

We plan to make money by selling access to the Platform in support of clinical trials for responder verses non-responder, researchers looking to develop testable hypotheses, or pharmaceutical companies looking to develop companion diagnostics. Access to the Platform will likely follow a COGS plus margin model on a per-sample basis. In addition, we plan to sell access to the Knowledgebase for commercial and non-commercial data mining applications via a subscription based model.



About our Competition

At the moment genomics (e.i. 23AndMe, Pathway Genomics) profiling represents the only standardized platform for clinical companion testing knowledgebase mining. Given that our proposal is to enter this space in a tangential field, we feel there is limited competition if not the inverse of complementing the genomics space.



Progress with Customers to date

Innovation Details

Intellectual Property Summary

The intellectual property behind Haystack is composed of several independent but connected pieces. Sample Processing is covered under the terms, conditions and licenses attributable to each individual chemical, component and material. Molecular Data Collection follows the guidelines established by the Human Proteome Organization, and while the conventions established under this organization are used in guidance, modifications and amendments are considered proprietary.  Data Characterization and Statistical Bioinformatics follow the basic tenants of label-free quantitation of complex proteomics with the exact processes and innovations considered proprietary. The Knowledgebase derived from this work will be leveraged for biomarker mining and discovery.

Clinical Information

Currently, LCMS-based proteomics has eluded clinical practice, yet the technique continues to show promise of clinical utility. Still considered within the discovery phase, LCMS-based proteomics can detect tens of thousands of potential biomarkers for cancer and other diseases, as reported in the literature. [1,2] A current search of literature will result in over 29,000 hits for mass-spectrometry-based protein or peptide biomarkers mentioned anywhere in the article.  Despite this abundance of literature, progress toward clinical applications faces two distinct hurdles, robustness and utility. [3] Anderson and coworkers succinctly outlined this pointing out the inverse relationship between an increase in identified plasma based protein biomarkers over time while there is an apparent decrease in FDA approved diagnostics. [4] Despite the depletion of low-hanging-fruit, several studies continue to indicate string clinical utility [5-9]. It stands to reason that formulating a standard, single derivative solution, addressing the largest sources of variability [10], will have future clinical efficacy.

 

  1. Van Eyk, J.E., et.al., Clinical Proteomics, John Wiley & Sons, 2008.
  2. Petricoin, E.F., et.al., Nat Rev Drug Discov (2002), 1, 683–695
  3. Parker, C.E., et.al., Analyst (2010), 135, 1830–17
  4. Anderson, N.L., et.al., Molecular & Cellular Proteomics (2002), 1, 845–867
  5. Lin, Y?W., et al., International Journal of Gynecological Cancer 16.3 (2006): 1216-1224.
  6. JJones et. al. Clinical Colorectal Cancer, 15.2, (2016), 186-194
  7. Dalrymple, Annette, et al., Journal of proteome research 6.7 (2007): 2833-2840.
  8. Pendyala, Gurudutt, et al., Journal of proteome research 9.1 (2009): 352-358.
  9. JBlume et. al. J. Applied Laboratory Medicine, (2016)
  10. Piehowski et al. J. Proteome Res. 12.5 (2013)

Regulatory Status

We currently do not have any FDA approval. The applications of this project are directed towards clinical research, not clinical testing, yet.

How we will use the funds raised

Funds collected from this project are intended to be used toward the establishment of an analytical laboratory, purchasing scientific equipment and instruments, creating the analytical compute infrastructure and supporting the staff scientists. This is the first phase of our development, building out the infrastructure. Our second phase will entail demonstration of the reduction to practice by either contracting through a CRO, Academic Institution, or private entity for IRB approved investigations. Our third phase of development will be to involve the public directly in our efforts to survey as many diseases and maladies as possible, in search of a signal for each.

Thank You

Our goal is to provide a universal tool to aid in physician diagnostics, either by monitoring patient recoveries, establishing companion diagnostics, discovering treatment responders verses non-responders, or ultimately tracking the molecular profiles of healthy individuals and providing early disease indicators. The grandeur of this project is to become the future of personal medicine – that singular test platform that unlocks the mysteries of our health, currently.

Supporters

Comments

Login to post your comment!
Click here to Login

Updates

    No updates found .

$ 50

pledged of $ 10,000 goal

  • 11 Days left

24

Interest
Score

0

Adoption
Score


Rewards All contributions are tax-deductible.