Mailing list
Capture and share the knowledge contained in your experimental data by building models tuned to your proprietary chemistry.
• Build models for individual chemical series or projects
• Generate ‘corporate’ models based on large compound property databases
• Iteratively improve models as more data is generated for your chemistry
For users who are not computational experts, the Auto-Modeler provides an automated process to ensure that your data has been thoroughly analyzed to generate a validated predictive model. To achieve this, the Auto-Modeler can:
• Automatically apply multiple, advanced modeling techniques to your data and identify the most appropriate model to use (see FAQ for details)
• Split your data set into sub-sets for training, validating and testing the model using an automatic clustering algorithm based on 2D structural fingerprints
• Use the library of molecular descriptors provided, including whole molecule descriptors (e.g. molecular weight, logP and polar surface area) and 2D structural descriptors
For expert users, facilities are provided to:
• Manually split your data set
• Input your own 2D descriptors (as SMARTS) or import values for other types of descriptor, including experimental data
• Choose the modeling technique(s) to apply
Models generated with Auto-Modeler benefit from StarDrop’s unique Glowing Molecule visualization, allowing you to intuitively make the link between molecules’ structures and their properties, highlight ‘problem’ areas of molecules and guide the design of molecules towards improved properties.
Download our Fequently Asked Questions document for further information about the Auto-Modeler.
