This option allows you to make your own kcc models either on your own data (local table) or from ChEMBL.

From your own data:

• Read in a chemical spreadsheet containing your chemical 2D structure (mol column) and an activity column containing data such as pKd, mM, uM or nM.
• Select Chemistry/MolScreen/ Make Chemical Classification Model
• Select the Local Table tab.
• Give your model a unique name.
• Find your chemical table from the drop down list.
• Find the column name containing activity data from the drop down list.
• Select the activity unit pKd, mM, uM or nM.
• Select whether you would want pKa charge prediction.
• Click OK.

From ChEMBL data:

• Give your model a unique name.
• Enter a Uniprot ID (e.g. 5ht1f)
• Select whether or not you want to include all mammalian data or just human.
• Check makemodel to make the model immeditately.

How to use the model to predict:

• When the model has been built you will see the model listed in the ICM workspace as shown below.

To run the model against a set of chemicals:

• Read in a chemical spreadsheet containing your chemical data.
• Right click on the model in the ICM workspace and choose Predict.
• Enter the table name and prediction model you would like to use.
• Once the prediction has finished you will see three additional columns: the kcc value, kca value and the tanimoto distance to the nearest compound in the model dataset.