--- title: "Readme" task: "" lineage_type: import upstream_source: https://github.com/PatWalters/resources_2025/blob/fdc9ddb3/README.md upstream_sha: fdc9ddb3 imported_at: 2026-06-26 prompt_class: catalogue upstream_changes: accepted author: upstream validated: false --- ## Machine Learning in Drug Discovery Resources 2025 ### Books [Drug Design: From Structure and Mode-of-Action to Rational Design Concepts](https://www.amazon.com/Drug-Design-Structure-Mode-Action/dp/3662689979) As cheminformatics practitioners, we need to understand the drug design process. This book, written by Prof. Gerhard Klebe, a pioneer in the field, provides an excellent overview of numerous drug design approaches. [Python for Data Analysis: Data Wrangling with pandas, NumPy, and Jupyter](https://www.amazon.com/Python-Data-Analysis-Wrangling-Jupyter/dp/109810403X) Programming and data science are critical elements of cheminformatics. This book, written by Wes McKinney, the author of the widely used Pandas library, provides a great starting point for learning Python and applying it in data science. [Data Science from Scratch: First Principles with Python](https://www.amazon.com/Data-Science-Scratch-Principles-Python/dp/1492041130/) This book provides another great introduction to data science. It provides an introduction to several critical topics, including Python, Statistics, Probability, Machine Learning, Clustering, and Databases. [Statistics in a Nutshell: A Desktop Quick Reference](https://www.amazon.com/Statistics-Nutshell-Desktop-Quick-Reference/dp/1449316824/) [Machine Learning with PyTorch and Scikit-Learn: Develop machine learning and deep learning models with Python](https://www.amazon.com/Machine-Learning-PyTorch-Scikit-Learn-learning/dp/1801819319/) To effectively apply cheminformatics, one needs a solid grasp of statistics. This book provides a good overview with code examples. [Machine Learning with PyTorch and Scikit-Learn: Develop machine learning and deep learning models with Python](https://www.amazon.com/Machine-Learning-PyTorch-Scikit-Learn-learning/dp/1801819319/) Machine learning (ML) has become an integral component on cheminformatics. This book provides a fantastic introduction to more traditional ML approaches and recent advances in deep learning. ### Datasets You'll notice the conspicuous absence of two widely used datasets, [MoleculeNet](https://moleculenet.org/) and the [Therapeutic Data Commons (TDC)](https://tdcommons.ai/), from this list. Both of these datasets are highly flawed and should not be used. For more on the reasons why, please consult this [blog post](https://practicalcheminformatics.blogspot.com/2023/08/we-need-better-benchmarks-for-machine.html). [OpenADMET](https://openadmet.org) seeks to proactively characterize the chemical space accessible to ADMET-associated proteins (“anti-targets”). By applying recent advances in experimental and computational techniques, a comprehensive open library of experimental and structural datasets will be generated. It's early days for OpenADMET, but knowing the folks involved, I'm highly optimistic. [AIRCHECK](https://aircheck.ai) is a platform that provides access to a large collection of high-quality datasets for drug discovery and development. The datasets are curated from various sources and are available in a standardized format. The current focus appears to be on DNA-encoded library (DEL) data. [Polaris](https://polarishub.io) aims to improve the state of benchmarking so ML can have a more significant impact on real-world drug discovery scenarios. To start, Polaris hopes to provide a single source of truth that aggregates and provides simple access to datasets & benchmarks. [PLINDER](https://plinder.sh) is an academic-industry collaboration to collect and organize protein-ligand interaction data. The effort is driven by VantAI, NVIDIA, the Computational Structural Biology group at the University of Basel & SIB Swiss Institute of Bioinformatics (co-organizers of CASP), and MIT. PLINDER aims to provide a gold standard dataset and evaluations to push the field of computational protein-ligand interactions prediction forward. ### Blogs [Eric J Ma's Website](https://ericmjl.github.io/) Eric's blog provides an excellent introduction to the application of cutting-edge informatics in drug discovery. [Oxford Protein Informatics Group (OPIG)](https://www.blopig.com/blog) This blog contains a lot of great [Bio|Chem]informatics content, chock-full of code. [Charlie’s Substack](https://harrisbio.substack.com/) Charlie Harris writes about applications of AI in drug discovery. Most recently, his posts have focused on efforts to reproduce AlphaFold3. [Mogan Thomas' Cheminformatics Blog](https://cheminformantics.blogspot.com/) This one is new, but it looks promising based on the first post. [Jon Swain's Blog](https://jonswain.github.io/) Jon Swain, a second-generation Cheinformatics blogger, has a great set of Jupyter notebooks demonstrating key concepts. [Practical Cheminformatics](https://practicalcheminformatics.blogspot.com/) This is a blog where I post once a month or so. These posts typically contain code demonstrating various aspects of cheminformatics; clustering, machine learning, data visualization, etc. I occasionally post opinions on things like AI and getting a job. [Is Life Worth Living](https://iwatobipen.wordpress.com/) A great blog from Iwatobipen (aka pen), whose posts are chock-full of great code examples. Pen always seems to be up on the latest methods and posts interesting examples on various topics ranging from quantum chemistry to machine learning. [The RDKit Blog](http://rdkit.blogspot.com/) Greg Landrum is the primary contributor to and BDFL of the RDKit. In addition to the latest and greatest features in the RDKit, Greg's posts also touch on a number of key issues in Cheminformatics, such as dealing with unbalanced datasets and the impact of fingerprint folding on similarity searching. [Models to molecules](https://driesvr.github.io/) A new blog by Dries Van Rompaey that is off to a great start. ### Tutorials [Practical Cheminformatics Tutorials](https://github.com/PatWalters/practical_cheminformatics_tutorials) I put together this collection of Jupyter notebooks to demonstrate various aspects of cheminformatics and machine learning. The notebooks illustrate a range of topics from cheminformatics basics to more advanced machine learning. The tutorials all use open source software and can run on Google Colab without installing software locally. [TeachOpenCADD](https://github.com/volkamerlab/TeachOpenCADD) A great set of tutorials from Andrea Volkamer's group that use open-source software to teach Computer-Aided Drug Design concepts, including molecular similarity, applications of machine learning, and pharmacophore analysis. [The RDKit Cookbook](https://www.rdkit.org/docs/Cookbook.html) A terrific resource that provides "recipes" for a number of common tasks. [Vina Colab Tutorials](https://autodock-vina.readthedocs.io/en/latest/colab_examples.html) A tutorial set shows how to run Autodock Vina and the associated protein and ligand setup utilities on Google Colab. [GNNs for Chemists](https://github.com/HFooladi/GNNs-For-Chemists) A great introduction to graph neural networks (GNNs) by Hosein Fooladi. [PDB-101 from the RCSB PDB](https://pdb101.rcsb.org/train/training-events) The Protein Databank (PDB) has a wide range of tutorials available. The Python scripting tutorials are very good.