Modeling and predicting drug-relevant properties of molecules is a critical issue for scientists working in drug discovery research.  Strand’s new generation tool, Sarchitect, is meant for Quantitative Structure Activity  Relationship (QSAR)  studies and modeling of ADME / Tox properties. Sarchitect allows model builders to build best possible models of their data and equally importantly, monitor and improve the models over a period of usage. It allows users of the models to optimize complex multi-dimensional ADME/Tox properties of molecules while retaining their biological activity.

The Sarchitect platform empowers computational and medicinal chemists, modelers, DMPK scientists and other users with:

• QSAR modeling and deployment platform.

• Powerful algorithms, interactive views, single click model building, intuitive workflows for chemists and customized scripting.

• Ready to use predictive QSAR models for ADMET end points.

• Successful drugs, virtually now!

Quick and cost-effective way to prioritize candidates in pipeline

Readily available models

• ADME

  • Bioavailability

  • Plasma Protein Binding

  • Volume of Distribution

  • Aqueous Solubility

  • Absorption

  • Elimination Half Life

  • Toxicity

• Toxicity

  • hERG channel binding

  • Mutagenicity

  • Carcinogenicity

Superior model-building capabilities

• Single click automated model building feature

• Building of ‘best possible models' by capturing ‘best available' practices

• Built in robust validation methods for models including Y-scrambling and descriptor sensitivity; single click external test validation

• Model building algorithms include Decision Trees, Naïve Bayes, Decision Forests, Multilinear Regression PLS, Neural Networks and Support Vector Machines.

• Easy interface to combine models to create ensembles and nested models.

 

Compound profiling based on multiple properties

Easy evaluation of applicability domains of models

Highly user-friendly and publication-friendly

• Publication quality images.

• Extremely  ‘easy-to-use’ features and workflows.

• Variety of data exploration methods.

• A comprehensive descriptor list including 2D, 3D and surface area

• Computation and visualization of similarity, on structural and descriptor spaces

• Feature selection methods include Genetic algorithm and wrappers based on forward selection and backward elimination

Cross validation in-built into building workflow