Any number of molecules, uploaded as a dataset or selected in a generator, can be scored using pre-defined scorers and saved in the form of a new Makya dataset.
Often QSAR models or models from an external API are used to guide molecule generation. However, in some cases, using these models during generation may not always be desired. Such cases include:
- chemical space exploration,
- models that struggle to be optimized together,
- etc.
Herein we will review how to score molecules in Makya using previously trained models (or descriptors), outside of a context of generation.
1) Set up a new scorer following the instructions contained within the article Setting up a new Scorer.
2) Once the desired parameters have been selected and the scorer has been saved, navigate to the tab "Datasets", and select the top right button Score a new dataset.
3) Within the "Create Dataset to Score" window, molecules to score can be imported as a SMILES string, drawn with the sketcher, imported as a CSV file, or copied from an existing dataset. Once a molecule or molecules have been selected click Save.
4) Molecules will then be re-scored and saved as a new Makya Dataset. The newly scored molecules can be located in the "Datasets" tab or by selecting Open dataset.