Online Course: [L4-CA] Machine Learning for Chemical Applications

Course Focus
Learn to build advanced cheminformatics workflows in an interactive hands-on setting using KNIME Analytics Platform and its Cheminformatics Extensions.

This online course will contain several units where we’ll cover a diverse set of topics such as data wrangling, interactive filtering of compounds and their exploration, machine learning for virtual screening, multi-objective optimization, and matched molecular pair analysis. After completing this course you’ll have a set of fully functional workflows and will have learned how to build your own advanced applications.

This is an instructor-led course consisting of four 75-minute online sessions, each session has an exercise for you to complete at home, we will go through the solution at the start of the following session. The course concludes with a 15-minute wrap up session.

Course content

  • Session 1: Access and Transformation of Chemical Data
  • Session 2: Machine Learning
  • Session 3: Multi-objective Optimization
  • Session 4: Matched Molecular Pair Analysis
  • Session 5: Discussion of the Last Exercises and Q&A

If you are interested in signing up:

Thank you for the course. It was very useful to learn about these techniques for chemicals.

I was wondering if the MMP analysis can be used to discover new structures with better activity or properties. For example, use MMP to create novel structures from the data we used to build the model, then predict with the model to find those with better activity/properties. Are there any example workflows of this using MMP or any other methods? I remember in the other course, Greg Landrum mentioned the ability to create new structures after R-group decomposition.

Hi @tnad

Thanks for your question!
What you proposed could be done, but I think we don’t have one single workflow in place that is covering all the bits you mentioned. There is a workflow that shows how to do R-group decomposition together with library enumeration, (and the same bits are also covered here), but you’d still need to combine it with the prediction model part.

Let me know in case that doesn’t answer your question :slight_smile:

Hello, I am a Knime beginner and learning to use Knime for chemical applications. I downloaded this workflow but found the original dataset is not available any more. Could you please update the link to the initial dataset?
Thank you very much!