About this add-in
Add-ins for Materials Informatics (MI) and Chemoinformatics. Materials Informatics is the effort to leverage information science techniques such as machine learning to make materials development and discovery more efficient.
The challenge in dealing with molecules using machine learning is how to convert molecules into numbers that a computer can interpret. One solution is the "fingerprint," which is a numerical representation of the structural characteristics of a compound, also known as a feature. Various descriptors have been proposed for such features.
This add-in uses a library called RDKit to calculate these descriptors from SMILES. The calculated descriptors can be used as explanatory variables in predictive models.
With this add-in, you can:
- Generate molecular descriptors
- Generate a fingerprint
- Drawing molecules
- Preprocessing Descriptors for Use in Predictive Models
For more information about the RDKit, see the following page.
https://www.rdkit.org/
System Requirements
- JMP 18 and later
- Display language is English and Japanese only
- Windows OS
- The following Python packages are installed:
- RDKit
- Numpy
- Pandas
Advance preparation
Installing the RDKit
- Select File > New > JSL Script.
- Enter the following script and select [Edit] > [Run Script].
Python Install Packages("numpy pandas rdkit");
How to use
Generation of molecular descriptors/fingerprints
- Select "Add-ins" > "Toolkit for Materials Informatics" > "Descriptor Calculation".
- Specify the column containing the SMILES as "Y,SMILES". Check the descriptors and fingerprints you want to generate from the method selection list.
- Click the "OK" button to generate the selected features.
- Click the "Close" button.
How to use this add-in
This add-in for materials informatics and chemical informatics. This add-in uses an RDKit(https://www.rdkit.org/ ) to calculate these descriptors from SMILES. The calculated descriptors can be used as feature variables in the prediction model.
Here's what you can do:
- Generate molecular descriptors
- Generate a fingerprint
- Draw molecules
- Preprocess descriptors for use in predictive models
You must install RDKit before using this add-in .
- Select File > New > JSL Script
- Enter the script below and select Edit > Run Script (JMP 18 required)
Python Install Packages("numpy pandas rdkit");
- Select Add-ins > Toolkit for Materials Informatics > Descriptor Calculation.
- Specify the column in which SMILES is entered as "Y,SMILES". Check the descriptors and fingerprints you want to generate from the list of method selections.
Excellent add-in @yuichi_katsumur ! Chemistry is arguably the most popular branch of science for JMP and you have just provided us easy access to the powerful and feature-rich RDKit package, making QSAR in JMP much more accessible.
The clear instructions for installing RDKit in Python are very easy to follow and really appreciate the nice checking utility with attractive output.
The Descriptors and three kinds of Fingerprints are great inputs to use for exploratory analyses (e.g. PCA, Multivariate Embedding, Hierarchical Clustering, Response Screening) and predictive modeling (e.g. Fit Model, Neural, SVM, XGBoost, Torch).
The molecule drawings are fast and make compelling labels in the JMP table.
The JSME editor and substructure search capabilities are remarkable and a very welcome bonus.
The attention to detail in this add-in is terrific! 
<- SMILE
