AVO (Autonomous Virtual Organism) employs contemporary computational techniques and machine learning to modernize drug discovery. AVO’s underpinnings come from structural biology and computational chemistry rather than systems biology. The platform accurately predicts activity against the target of interest, selectivity and drug-like properties (ADME) simultaneously, transforming discovery into a parallel process in a virtual environment, rather than a sequential process in a laboratory environment.


AVO applies virtual medicinal chemistry to optimize leads that result from an initial screen. New molecules are designed based on the drug-like leads and evaluated in AVO. AVO algorithms can predict which chemical changes provide an advantage and prioritize the molecules to be synthesized. A few rapid cycles of intensive in silico compound optimization, synthesis and testing replace the many cycles of synthesis and testing required by the traditional drug discovery processes. The AVO paradigm, therefore, requires far less time and money. AVO has a dramatic economic impact on the drug discovery process. Reduction in cycle time and cost through discovery and lead optimization can lower the capitalized cost by 63%, based on Evince’s analysis of a pharmaceutical productivity study published in Nature Reviews (March 2010). For a biotech company, lower cost enables prosecution of more molecules in parallel and higher probability of success.

AVO logo



AVO Diagram

Machine learning is only as good as its input data

The application of artificial intelligence without high-quality relevant molecular descriptors is therefore useless. For this reason, Evince has focused substantial effort on a process to create fingerprints that have enormous predictive value with respect to interaction of molecules with complex biological systems. Each molecule is subjected to a carefully calibrated series of biological simulations that generates a set of descriptors for the molecule, or a “biological fingerprint.” The approach used to describe molecules is critical for accurate and useful drug discovery and optimization.



AVO Diagram

Machine learning is unequaled in predicting properties from training data.

Machine learning is applied to interpret the descriptors and make accurate predictions on activity against the target, selectivity based on counter-screens and drug–like properties based on ADME filters. The AVO platform is unmatched in its accuracy and all-encompassing coverage of the drug discovery and optimization process.



AVO Diagram

AVO applies virtual medicinal chemistry to optimize leads

AVO optimizes leads that result from an initial screen. New molecules are designed based on the drug-like leads and evaluated in AVO. AVO algorithms can predict which chemical changes provide an advantage and prioritize the molecules to be synthesized. A few rapid cycles of intensive in silico compound optimization, synthesis and testing replace the many cycles of synthesis and testing required by the traditional drug discovery processes. The AVO paradigm, therefore, requires far less time and money to create clinical candidates.




Don't Compromise: AVO achieves unparalleled accuracy across all relevant properties - simultaneously.

  • Activity against target
  • Counter Screens for Selectivity
  • Activity in cellular models
  • Absorption, distribution, metabolism and excretion (ADME)
  • Toxicology
  • Multiple targets within the same pathway
  • Results generalize into new chemical space (novel scaffolds)