28 October 2011
Optibrium has launched StarDrop 5.1, an upgraded version of its support tool that helps drug discovery scientists to guide key decisions in drug discovery and quickly achieve successful project outcomes.
As a result of this upgrade, StarDrop’s capabilities to guide the design and selection of high quality, novel compounds are now available on the Apple Mac, a reflection of the increasing popularity of Macs within the industry.
StarDrop 5.1 is a software package that combines predictive capabilities with intuitive approaches to quickly and confidently target compounds with a good balance of properties, thereby reducing wasted effort and speeding progress to identify effective lead and candidate drugs. StarDrop’s unique approach to ‘multi-parameter optimisation’ explicitly accounts for the uncertainty in drug discovery data, whether due to experimental variability or predictive error, to provide scientists with a rigorous, objective analysis on which to make rational decisions.
As well as compatibility with the Apple Mac, StarDrop 5.1 brings a number of significant enhancements to existing features of StarDrop. For example, there is a new method for creating ‘chemical space’ visualisations based on the latest machine learning algorithm for ‘visual clustering’, that helps to easily explore the diversity of a project’s chemistry and identify ‘hot spots’ of high quality chemistry for further investigation.
There are also improvements to StarDrop’s plug-in modules: The Auto-Modeller sees the addition of the ‘Random Forests’ technique to its extensive repertoire for building predictive models tailored to a project’s chemistry and data. The addition of this new method has also led to improved QSAR models in the ADME QSAR module, which provides predictions of key ADME properties. StarDrop’s P450 metabolism module has been upgraded to provide improved prediction of regioselectivity and lability of metabolism by Cytochrome P450 enzymes, specifically around N-oxidation pathways. Finally, in response to user feedback on the new Nova module, further advances allow the generation of novel, relevant compounds ideas, prioritised against a project’s required property profile, starting from multiple initial structures and filtered according to user-specified rules to avoid unwanted substructures.