This overview gives a quick introduction to all StarDrop's standard and optional features
Probabilistic Scoring
Compare all your candidate molecules against a weighted profile of required properties to quickly see which are most and least likely to succeed.
- Allow for uncertainty: confidence in each score is calculated to highlight statistically significant compounds
- Weight properties: allow for trade-offs between properties for your particular project
- Use data from any source: StarDrop ADME QSAR or P450 models or imported in vitro, in vivo or in silico data
Chemical Space & Compound Selection
Visualise property trends across the chemical diversity of your compounds. Combine this with the compound selection tool to quickly explore the impact of different compound selection strategies in silico before carrying out more expensive in vitro and in vivo exploration.
- Multiple data sets: compare different chemical series and thousands of data points on one plot
- See the spread of properties across your project space
- Quality vs. diversity: Quickly explore and visualise the trade-offs between focused and diverse selections
Glowing Molecule
The Glowing Molecule visualisation shows you the regions of candidate molecules which may have the most influence on predicted properties. Try out new ideas interactively with instant feedback.
- Edit your compound: Explore the impact of different designs by seeing the properties and scores change as you edit
- Link structure and properties: easily visualise the influential regions of your molecules
- No more black-box models! Interpret predictions easily, even if the underlying models are nonlinear
Optional Features
ADME QSAR Module
The ADME QSAR module enables you to predict a broad range of ADME and physicochemical properties, prior to synthesis, using a suite of high-quality models, including:
|
|
P450 Module
The P450 metabolism models identify the regions of a candidate molecule most vulnerable to metabolism by the three major cytochrome P450 enzymes. These enzymes account for around 95% of human Phase I metabolism. Using StarDrop's P450 metabolism models, molecules can be redesigned to overcome possible metabolism problems or drug-drug interactions. Combining these models with other factors such as the molecule's affinity or lipophilicity can demonstrate the risk of high metabolic turnover.
Auto-Modeler Module
The Auto-Modeler gives both novice and expert users access to the tools needed to thoroughly analyze data and produce validated, predictive models. Output from the Auto-Modeler includes Glowing Molecule. For non-expert users, the Auto-Modeler can:
- Automatically generate models: multiple advanced modelling techniques, including Gaussian Processes, Radial Basis Functions, PLS and Decision Trees can be applied to data, helping you identify the most appropriate model
- Train, test and validate: automatically split your data into sub-sets to rigorously choose and validate your models
- Generate descriptors: MWt, logP, polar surface area and many other 2D structural descriptors are included with the Auto-Modeler
Cross Platform Integration
StarDrop's data integration capabilities mean that your project scientists can be provided with direct access to in-house models, databases and algorithms through StarDrop's intuitive, user friendly interface. Existing models and data analysis algorithms can be integrated using customizable Python scripts and interfaces to Accelrys' Pipeline Pilot.
