Reproducible biological fingerprinting enables Machine Learning driven safety predictions . For each new compound we screen, a standardized, digital cell based profile is created. This profile, or feature rich biological fingerprint, encodes the specific physiological or toxic effect on critical indicators of cellular stress for each unique compound.
Our focus is providing data quality , not quantity, as reproducibility and accuracy of any cell based phenotypic screens are critical to enabling the predictive potential of machine learning.
The digital biological fingerprint for each new compound can then be compared to a library of similar digital fingerprints from a carefully selected training set of marketed therapeutics, discontinued and withdrawn drugs and known toxins. Based on similarity, we can provide chemists with a calculated prediction of the potential safety / toxicity risk for any new compound we screen.
To continuously enable our clients to improve the fidelity of their risk predictions, our database houses an ever expanding library of digital biological fingerprints for comparison, representing thousands of functionally and structurally diverse compounds from across the broader chemical producing industries – Pharmaceutical, Cosmetics, Industrial and Agricultural chemicals. This enables any size company, from across all industry segments, to directly compare and determine the safety risk in early research for compounds in a new chemical series
Once our Machine Learning algorithms have compared the similarity of the fingerprint for a new compound to similar fingerprints from the broader compound training set in our database, the nearest neighbor similarity of cellular stress phenotypes can then be used to predict risk to human safety. The results are then color coded to enable simple visual interpretation of risk, allowing for rapid pipeline prioritization of compounds to be Approved, Improved or Removed at the Hit-to-Lead stage and through lead optimization.

We color code the risk to make it easier for chemists to make informed decisions on which compounds to approve, improve or remove from the Discovery pipeline which in turn will help in reducing the massive costs associated with downstream clinical trial failures or post-market withdrawals. Our entire compound library will be available via a cloud based subscription