7.5.4. SVM Classification Models

In this tutorial, you will build Support Vector Machines (SVM) classification models with a data set that is stored in the ..\Tutorial\Example3 folder. The descriptor values in each record do not relate to any specific chemical structure, so the same arbitrary structure - methane - has been added for each row. The data set contains 215 records, with each record containing a categorical output (an important ADMET property) and 171 pre-generated descriptors.

NOTE:

You can use this trick – inserting arbitrary SMILES structure(s) as the first column in Excel – to convert any existing tabular data in Excel into a SMILES input file recognizable by ADMET Predictor. The first row should contain column titles; “SMILES” being the first one. Just make sure to export the resulting Excel spreadsheet as tab-delimited text with .smi extension (you may have to rename it in Windows Explorer since Excel inserts .txt by default). In addition, if any strings in your spreadsheet contain commas (e.g., IPUAC names of chemicals) then Excel will automatically wrap them in double quotes. Oh well, one of the Excel’s little annoyances – you may want to edit the .smi file in a text editor to replace all double quotes with an empty string.