LiveDesign ML (Beta) Panel

Predict molecular properties with LiveDesign ML models and manage model training datasets.

To display this panel, do one of the following:
  • click the Tasks button and browse to LiveDesign ML Modeling → LiveDesign ML (Beta)
  • Using
  • Features
  • Additional Resources

Using the LiveDesign ML (Beta) Panel

Use this panel to predict molecular properties with trained ML models or to upload training data for model datasets. Before using this panel, you must connect to a LiveDesign instance that has LiveDesign ML enabled. If Maestro is not connected, the Connect to LiveDesign Dialog Box is displayed when the panel opens. The connected LiveDesign instance is shown in the panel title bar. For more information on LiveDesign ML, see the LiveDesign ML User Manual.

Predicting Properties

To predict molecular properties using a LiveDesign ML model:

  1. Select Predict property.

  2. Choose the source for the input structures and load them.

  3. Optionally, select a LiveDesign project to filter the available model datasets.

  4. Select the model dataset to use for prediction.

  5. Click Predict.

The prediction job runs on the LiveDesign ML server and can be monitored in the Job Monitor panel. If the connection to LiveDesign is interrupted or you close Maestro while a job is running, Maestro resumes checking the job's status automatically the next time you open the project and reconnect to LiveDesign. When structures are sourced from the Workspace or Project Table, the results are incorporated into the Project Table as two new properties: ml prediction-dataset and ml uncertainty-dataset. When structures are sourced from a file, the results are written to an output file and a notification banner displays the file path. Structures with 500 or more atoms are skipped during prediction. A maximum of 10,000 structures can be submitted per prediction job.

Uploading Training Data

To upload training data to create or update a model dataset in LiveDesign ML:

  1. Select Upload data.

  2. Choose the source for the structures. When File is selected, CSV dataset files can also be loaded directly.

  3. Select the property to associate with the training data.

  4. Select New dataset to create a new dataset, or Existing dataset to update an existing one.

  5. For a new dataset, select the LiveDesign project and enter a name. For an existing dataset, select the project to filter the available datasets, then select the dataset to update.

  6. Click Create or Update.

A notification banner confirms the outcome.

LiveDesign ML (Beta) Panel Features

Choose task options

Select the task to perform. Predict property generates property predictions for structures using a LiveDesign ML model. Upload data uploads training data to create or update a model dataset.

Predict property option and section

Select to predict molecular properties for structures using a LiveDesign ML model dataset.

Get structures from option menu

Choose the structure source for the current task.

  • Project Table (n selected entries)—Use the entries that are currently selected in the Project Table or Entry List. The number of entries selected is shown on the menu item. An icon is displayed to the right which you can click to open the Project Table and select entries. When this option is selected, a Load button is displayed to the right.
  • Workspace (included entries)—Use the entries that are currently included in the Workspace, treated as separate structures.
  • File—Use the specified file. When this option is selected, the File name text box and Browse button are displayed.
Select LiveDesign project option menu

Select a LiveDesign project to filter the available model datasets. If a dataset is selected directly without first selecting a project, this option menu is automatically updated to show the corresponding project.

Select model dataset option menu

Select the model dataset to use for prediction. When a versioned dataset is selected, a version selector appears to the right. Metadata for the selected dataset, including the associated LiveReport name and last trained date, is displayed below.

Predict button

Click to submit the prediction job to LiveDesign ML. The job can be monitored in the Job Monitor panel. When structures are sourced from the Workspace or Project Table, the results are incorporated into the Project Table. When structures are sourced from a file, the results are written to an output file.

Upload data option and section

Select to upload training data to create a new dataset or update an existing one in LiveDesign ML.

Get structures from option menu

Choose the source of structures to upload: Workspace (included entries), Workspace (selected entries), or File. When File is selected, CSV dataset files can also be selected for direct upload.

Select property option menu

Select the structure property to associate with the training data. The menu lists properties common to all loaded structures. A Project Table button is displayed next to the option menu when structures are loaded from the Workspace or Project Table. Not available when a CSV dataset file is loaded.

Upload training data to options

Select New dataset to create a new dataset, or Existing dataset to update an existing one.

Select LiveDesign project option menu

Select the LiveDesign project for the dataset. When Existing dataset is selected, if a dataset is selected directly without first selecting a project, this option menu is automatically updated to show the corresponding project.

Enter new dataset name text box

Specify the name for the new dataset. Only present when New dataset is selected.

Select model dataset option menu

Select the existing dataset to update. Metadata for the selected dataset is displayed below. Only present when Existing dataset is selected.

Create button

Click to create the new dataset or update the existing one in LiveDesign ML. The button is labeled Create when New dataset is selected, and Update when Existing dataset is selected. A notification banner confirms the outcome.

Reset button

Reset the panel to its default settings.

Related Topics