How-to: Use Built Machine Learning Model for Property Prediction/Verification of Unseen Molecules

OpenEye Machine Learning floe build machine learning models that predict physical properties of small molecules. In this guide, a previously built and trained fully connected neural network model will be used to predict molecular solubility. First, find the Physical Property Prediction for Small Molecule using Machine Learning Floe in Orion. There are general directions for finding floes in the User Interface Guide.

Next, click on the barchart icon to the right of the floe title, so that the floe appears on the Analyze Enabled list.

Analye Enable

Let’s select a previously built neural network model for property prediction. This model has to be built using the Neural Network Based Regression Model Building floe. Using the table and model analysis from the floe report, a well fitted model can be chosen (Refer to Previous How-to on model optimization). For this guide, assume that the second model is best for our need, we need to note down the Record number #, 29 in this case.

Find Model

Now on the floe page, go to the ‘Show in Project Data’ and activate the Output Model.

Find Model
Find Model

Once active, models can be found in the analyze page. Select the required model, i.e. #29.

This selected model will be sent to the Property Predictor floe which has already activated.

Find Model
Find Model

Once the model has been sent to the field “Input tensorflow Model”, make sure the model ID is set propely to the ID you chose in the analyze page. This ID can be set in the “Model ID of which Tensorflow model to use to predict.: ” under the “Machine Learning Model Options” field.

Next, add in the small molecule dataset whose property needs to be predicted.

Note

Sometimes the model is sent to the wrong input (small molecule) instead of the Tensorflow model input. This error is shown in the screenshot below. Make sure you have model #29 from analyze page in the Tensorflow model input and a dataset of small molecules in the first input.

Find Model

If the molecule dataset is used as validation, and already has said properties precalculated, then select the appropriate column in ‘Validation Field’ and the floe will produce R2 and other measures between the prediction and baseline.

Find Model
  • Finally, in the field Model ID of which Tensorflow model to use to predict enter 29. This ID should match the ID in your model.

That’s it! Run the floe.

Note

The output floeport will look very similar to this Floereport and Analyze.

Library Details of the Floe