Practical Cheminformatics Index

16 minute read

Published: April 12, 2026

Unified index of posts from Blogger and GitHub Pages.

Total posts: 87

Generated on 2026-04-11.

Date	Title	Source	Keywords	Description
2026-02-07	AI in Drug Discovery - Please Stop Fishing in the Bathtub!	GitHub	AI, Drug Discovery, Benchmarking	Critiques the misuse of the DUD-E dataset for validating machine learning models, arguing that models often “cheat” by recognizing simple chemical patterns rather than actual protein-ligand interactions.
2026-01-17	Practical Cheminformatics with Marimo	GitHub	Marimo, Jupyter, Python, Visualization	Introduces Marimo, a reactive Python notebook, as a superior alternative to Jupyter for cheminformatics with features like reactive execution and interactive UI elements.
2026-01-09	Can Gemini Search the ChEMBL Database?	GitHub	Gemini, LLM, ChEMBL, SQL	Evaluates the ability of the Gemini LLM to generate SQL queries for the ChEMBL database, noting that while it succeeds at query construction, data quality remains a limiting factor.
2025-11-08	Performing Exploratory Data Analysis on the OpenADMET ExpansionRx Blind Challenge Dataset	GitHub	EDA, OpenADMET, ADMET, Data Analysis	Provides a guide and notebook for performing EDA on the OpenADMET ExpansionRx dataset, emphasizing the importance of understanding data before modeling.
2025-11-03	We Still Haven’t Found What We’re Looking For - The Continuing Evolution of Protein-Ligand Co-Folding Methods	GitHub	Cofolding, Protein-Ligand, AlphaFold	Discusses the evolution of protein-ligand co-folding methods like PEARL and OpenFold3, noting their ongoing struggles with novel ligands and binding pockets.
2025-09-20	Just Because You Published It Doesn’t Mean It’s Right	GitHub	Reproducibility, Publication, Benchmarking	Critiques aqueous solubility modeling papers that use misleadingly “easy” benchmarks and advocates for higher quality, certified datasets.
2025-09-15	Time For a New Adventure	GitHub	Career	Announces the author’s new role as Chief Scientist at OpenADMET and outlines the initiative’s mission to improve drug metabolism and toxicity prediction through open science.
2025-07-27	Redoing the Boltz-1 Analysis of Orthosteric and Allosteric Ligand Cofolding with Boltz-2	GitHub	Boltz-1, Boltz-2, Cofolding, Allosteric	Updates a previous analysis with the Boltz-2 model, concluding that allosteric pose prediction remains a significant challenge despite model improvements.
2025-07-21	Three Papers Demonstrating That Cofolding Still Has a Ways to Go	GitHub	Cofolding, Protein-Ligand	Summarizes three studies highlighting the limitations of current co-folding models, including their reliance on training set similarity and failure to identify allosteric sites.
2025-05-18	GNN’s can extrapolate for some properties, but there’s a trick	GitHub	GNN, Machine Learning, Extrapolation	Explains how Graph Neural Networks (GNNs) can extrapolate properties like molecular weight using “norm aggregation,” correcting a previous finding.
2025-05-12	Useful RDKit Utils - A Mötley Collection of Helpful Routines	GitHub	RDKit, Python, Utilities	Provides a collection of helpful Python routines for common cheminformatics tasks using the RDKit library.
2025-05-06	The Trouble With Tautomers	GitHub	Tautomers, RDKit, Cheminformatics	Discusses the challenges of handling tautomers in cheminformatics and demonstrates how to use RDKit for tautomer enumeration and standardization.
2025-04-26	Why Don’t Machine Learning Models Extrapolate?	GitHub	Machine Learning, Extrapolation	Explores the fundamental reasons why traditional machine learning models often fail to generalize beyond the chemical space of their training data.
2025-04-26	We’ve Moved	Blogger	General	Announces the blog’s move from Blogger to a new Markdown-based site on GitHub Pages.
2025-03-07	Even More Thoughts on ML Method Comparisons	Blogger	Machine Learning, Benchmarking	Emphasizes the need for rigorous statistical testing and better visualization when comparing machine learning models in drug discovery.
2024-11-16	Some Thoughts on Splitting Chemical Datasets	Blogger	Data Splitting, Machine Learning, Validation	Examines how different data-splitting strategies (Random, Scaffold, Butina, UMAP) affect the evaluation and generalization of machine learning models.
2024-10-08	Silly Things Large Language Models Do With Molecules	Blogger	LLM, SMILES, Cheminformatics	Critiques the use of general-purpose LLMs for molecule generation, showing they often rely on simple string manipulation rather than chemical understanding.
2024-09-30	Digging Deeper into Thompson Sampling - A Guest Blog Post by Patrick Riley	Blogger	Thompson Sampling, Active Learning, Optimization	Analyzes why certain search tasks, like docking, are significantly more difficult to optimize using Thompson Sampling than others.
2024-05-22	Generative Molecular Design Isn’t As Easy As People Make It Look	Blogger	Generative Models, Molecular Design	Discusses the extensive “child-proofing” and post-processing required to make generative AI models produce physically valid and chemically stable molecules.
2024-01-16	AI in Drug Discovery - A Highly Opinionated Literature Review (Part III)	Blogger	AI, Drug Discovery, Literature Review	Provides a curated list of review articles from 2023 covering property prediction, docking, and multi-objective optimization.
2024-01-08	AI in Drug Discovery - A Highly Opinionated Literature Review (Part II)	Blogger	AI, Drug Discovery, Literature Review	Reviews 2023 developments in LLMs, Active Learning, and Explainable AI within the context of drug discovery.
2024-01-02	AI in Drug Discovery 2023 - A Highly Opinionated Literature Review (Part I)	Blogger	AI, Drug Discovery, Literature Review	Focuses on 2023 advancements and critiques in docking, AlphaFold2, and benchmarking methodologies.
2023-12-07	Some Thoughts on Biotech vs Pharma for Computational Chemists	Blogger	Career, Biotech, Pharma	Compares career paths in biotech and pharma, discussing mentorship, software access, and the challenges of managing AI hype.
2023-11-27	Comparing Classification Models - You’re Probably Doing It Wrong	Blogger	Machine Learning, Classification, Metrics	Critiques common mistakes in comparing classification models and suggests more robust metrics and validation strategies.
2023-08-03	We Need Better Benchmarks for Machine Learning in Drug Discovery	Blogger	Benchmarking, Machine Learning, Drug Discovery	Advocates for the development of more realistic and challenging benchmarks that reflect the complexities of actual drug discovery projects.
2023-07-05	A Simple Tool for Exploring Structural Alerts	Blogger	Structural Alerts, Toxicity, Screening	Introduces a tool for identifying, visualizing, and understanding structural alerts to aid in toxicity and screening assessments.
2023-06-12	Getting Real with Molecular Property Prediction	Blogger	Property Prediction, Machine Learning	Discusses the practical challenges of predicting molecular properties and the importance of data quality and model interpretability.
2023-05-02	Using Counterfactuals to Understand Machine Learning Models	Blogger	XAI, Counterfactuals, Machine Learning	Explores the use of counterfactual explanations to gain insights into how machine learning models make decisions about molecules.
2023-04-10	Build a QSAR Model in 8 Lines of Python	Blogger	QSAR, Python, Machine Learning	Demonstrates the simplicity and power of modern Python tools for building predictive QSAR models with minimal code.
2023-04-03	Getting Inside the Mind of the Medicinal Chemist with Machine Learning	Blogger	Medicinal Chemistry, Machine Learning, SAR	Investigates how machine learning can be used to capture and augment the intuition of experienced medicinal chemists.
2023-03-12	Working With Drug Data from the ChEMBL Database	Blogger	ChEMBL, SQL, Data Mining	Provides a practical guide to querying and processing drug-related data from the ChEMBL database using SQL and Python.
2023-02-01	Generative Molecular Design - We Need to Raise the Bar	Blogger	Generative Models, Molecular Design	Calls for more rigorous evaluation and higher standards for generative molecular design methods to ensure they provide real value.
2023-01-03	AI in Drug Discovery 2022 - A Highly Opinionated Literature Review	Blogger	AI, Drug Discovery, Literature Review	A comprehensive review of the most significant papers and trends in AI for drug discovery published throughout 2022.
2022-12-04	Mining Ring Systems in Molecules for Fun and Profit	Blogger	Ring Systems, Data Mining, RDKit	Explores different methods for identifying and analyzing ring systems in large chemical datasets using RDKit.
2022-03-28	Clustering Fragment Screening Hits With a Self-Organizing Map	Blogger	Clustering, SOM, Fragment-based Design	Demonstrates the use of Self-Organizing Maps (SOMs) to organize and analyze results from fragment-based screening campaigns.
2022-01-17	The Solubility Forecast Index	Blogger	Solubility, Physicochemical Properties	Explains the development and utility of the Solubility Forecast Index (SFI) in early-stage drug discovery.
2022-01-03	Useful RDKit Utilities	Blogger	RDKit, Python, Utilities	A collection of practical RDKit-based utilities for common tasks like molecule cleaning, standardization, and property calculation.
2021-11-30	Picking the Highest Scoring Molecule(s) From Each Cluster	Blogger	Clustering, Selection, Virtual Screening	Provides a Python-based workflow for selecting diverse, high-scoring molecules from virtual screening results using clustering.
2021-10-24	Exploratory Data Analysis With mols2grid and Bemis-Murcko Frameworks	Blogger	EDA, mols2grid, Murcko Frameworks	Shows how to perform interactive EDA using the mols2grid library and scaffold analysis with Bemis-Murcko frameworks.
2021-09-12	Similarity Search and Some Cool Pandas Tricks	Blogger	Similarity Search, Pandas, Python	Combines chemical similarity searching with advanced Pandas techniques to efficiently process and analyze molecule data.
2021-08-31	Building a multiclass classification model	Blogger	Machine Learning, Classification, Multiclass	Walks through the process of building and evaluating a multiclass classification model for chemical properties or activities.
2021-08-21	Practical Cheminformatics - The Directory	Blogger	Directory, Index	An annotated guide and directory to the various resources and posts available on the Practical Cheminformatics blog.
2021-07-27	Viewing Clustered Chemical Structures in a Jupyter Notebook	Blogger	Visualization, Clustering, Jupyter	Demonstrates techniques for visualizing and exploring the results of chemical clustering directly within Jupyter notebooks.
2021-07-07	Automatic Analog Generation With Common R-group Replacements	Blogger	Analog Generation, R-groups, SAR	Presents a tool for automatically generating chemical analogs by performing common R-group substitutions on a lead scaffold.
2021-06-03	Assessing Interpretable Models	Blogger	XAI, Interpretable ML	Discusses methods for assessing the interpretability of machine learning models and why it matters in drug discovery.
2021-03-30	Fast Parallel Cheminformatics Workflows With Dask	Blogger	Dask, Parallel Processing, Python	Explains how to use the Dask library to parallelize and accelerate large-scale cheminformatics workflows in Python.
2021-01-18	AI in Drug Discovery 2020 - A Highly Opinionated Literature Review	Blogger	AI, Drug Discovery, Literature Review	A critical review of the key developments and publications in AI for drug discovery during the year 2020.
2020-11-17	A Highly Opinionated List of Open Source Cheminformatics Resources	Blogger	Resources, Open Source	A curated and annotated list of essential open-source software, libraries, and datasets for cheminformatics.
2020-10-31	What Do Molecules That Look LIke This Tend To Do?	Blogger	SAR, Similarity, Data Mining	Uses similarity search and data mining to identify the typical biological activities associated with specific chemical motifs.
2020-10-12	A Collection of Things I Frequently Forget How To Do With Seaborn Scatterplots	Blogger	Visualization, Seaborn, Python	A handy cheat sheet for advanced formatting and customization of scatterplots using the Seaborn library.
2020-08-16	Examining the Data From the ChEMBL SARS-CoV-2 Drug Repurposing Screens	Blogger	ChEMBL, SARS-CoV-2, Drug Repurposing	Analyzes early drug repurposing screening data for SARS-CoV-2 released by the ChEMBL database.
2020-06-22	Wicked Fast Cheminformatics with NVIDIA RAPIDS	Blogger	RAPIDS, GPU, Parallel Processing	Demonstrates how to use NVIDIA RAPIDS to significantly speed up cheminformatics tasks using GPU acceleration.
2020-05-24	Using the Structure-Activity Landscape Index (SALI) to Analyze Data From the SARS-CoV-2 MPro Screen	Blogger	SALI, SAR, SARS-CoV-2	Applies the SALI index to identify activity cliffs and key SAR trends in SARS-CoV-2 MPro screening data.
2020-05-13	Some Thoughts on Comparing Classification Models	Blogger	Machine Learning, Classification, Metrics	Discusses the nuances of evaluating and comparing different classification algorithms for chemical datasets.
2020-05-04	Exploring the SARS-CoV-2 Main Protease (MPro) Structures	Blogger	SARS-CoV-2, MPro, Protein Structure	Provides an analysis of the first released crystal structures of the SARS-CoV-2 main protease.
2020-04-27	Positional Analogue Scanning	Blogger	SAR, Analogue Scanning, RDKit	Introduces a systematic approach for exploring the impact of substituent position on biological activity using RDKit.
2020-04-11	Adding Chemical Structures to a Recent COVID-19 Drug Repurposing Dataset	Blogger	COVID-19, Data Cleaning, SMILES	Documents the process of cleaning and augmenting a public COVID-19 dataset with proper chemical structures.
2020-03-30	Building on the Fragments From the Diamond/XChem SARS-CoV-2 Main Protease (MPro) Fragment Screen (Part II) Structure-Base Evaluation of Expanded Fragments	Blogger	Fragment-based Design, MPro, Structure-based Design	Evaluates fragment expansions for the SARS-CoV-2 main protease using structure-based modeling and docking.
2020-03-25	Building on the Fragments From the Diamond/XChem SARS-CoV-2 Main Protease (MPro) Fragment Screen (Part I)	Blogger	Fragment-based Design, MPro, SARS-CoV-2	Discusses initial strategies for growing and merging fragment hits identified from a large-scale screen against the SARS-CoV-2 MPro.
2020-03-21	Benchmarking “One Molecular Fingerprint to Rule Them All”	Blogger	Fingerprints, Benchmarking	Benchmarks a variety of molecular fingerprints to see which perform best across a range of common cheminformatics tasks.
2020-02-09	How (Not) to Get a Job in Science - Part 2 - The Interview	Blogger	Career, Interviewing	Offers practical advice and common pitfalls to avoid during the scientific job interview process.
2020-01-21	How to (Not) Get a Job in Science	Blogger	Career, Job Search	Shares insights and tips on navigating the scientific job market, from applications to networking.
2020-01-07	Visualizing Decision Trees	Blogger	Visualization, Decision Trees, Machine Learning	Explains how to create clear and informative visualizations of decision tree models used for chemical classification.
2019-11-13	Interactive Plots with Chemical Structures	Blogger	Visualization, Interactive Plots, Bokeh	Shows how to build interactive data visualizations that display chemical structures on hover using Bokeh.
2019-11-01	Visualizing Chemical Space	Blogger	Chemical Space, Visualization, Dimensionality Reduction	Explores different dimensionality reduction techniques for visualizing the chemical space of large molecular libraries.
2019-09-19	Dissecting the Hype With Cheminformatics	Blogger	Hype, AI, Cheminformatics	Uses cheminformatics analysis to critically examine the hype surrounding AI in drug discovery.
2019-07-28	How Good Could (Should) My Models Be?	Blogger	Machine Learning, Performance, Error Bar	Discusses the theoretical and practical limits of model performance based on experimental data uncertainty.
2019-06-02	Using Reaction Transforms to Understand SAR	Blogger	SAR, Reaction Transforms, RDKit	Demonstrates the use of RDKit reaction transforms to explore and understand structure-activity relationships.
2019-05-03	Where’s the code?	Blogger	Open Source, GitHub	Discusses the importance of code sharing in science and provides links to the blog’s open-source repositories.
2019-04-22	Clustering 2.1 Million Compounds for $5 With a Little Help From Amazon & Facebook	Blogger	Clustering, AWS, Large Datasets	Shows how to perform large-scale chemical clustering affordably using cloud computing and efficient algorithms.
2019-03-31	Multiple Comparisons, Non-Parametric Statistics, and Post-Hoc Tests	Blogger	Statistics, Multiple Comparisons	A practical guide to using robust statistical methods when comparing multiple experimental groups or models.
2019-03-03	Plotting Distributions	Blogger	Visualization, Statistics, Seaborn	Explains best practices for visualizing the distributions of chemical properties and experimental data.
2019-02-19	Some Thoughts on Evaluating Predictive Models	Blogger	Machine Learning, Evaluation, Metrics	Discusses the selection of appropriate metrics for evaluating predictive models in different drug discovery contexts.
2019-01-17	My Response to Peter Kenny’s Comments on “AI in Drug Discovery - A Practical View From the Trenches”	Blogger	AI, Drug Discovery, Debate	A formal response to technical critiques regarding the author’s perspectives on AI in drug discovery.
2019-01-11	K-means Clustering	Blogger	Clustering, Machine Learning	Provides a basic introduction and implementation of K-means clustering for chemical data.
2018-11-16	AI in Drug Discovery - A Practical View From the Trenches	Blogger	AI, Drug Discovery, Implementation	Shares practical insights and lessons learned from implementing AI solutions in real-world drug discovery projects.
2018-10-30	Self-Organizing Maps - 90s Fad or Useful Tool? (Part 1)	Blogger	SOM, Clustering, Machine Learning	Re-evaluates the utility of Self-Organizing Maps (SOMs) for modern cheminformatics and data visualization tasks.
2018-10-30	Self-Organizing Maps - The Code (Part 2)	Blogger	SOM, Python, Implementation	Provides a step-by-step Python implementation of Self-Organizing Maps for chemical data analysis.
2018-10-06	My Science/Programming Journey	Blogger	Career, Personal	A personal reflection on the author’s career path at the intersection of science and programming.
2018-09-29	Assigning Bond Orders to PDB Ligands - The Easy Way	Blogger	PDB, Bond Orders, RDKit	Demonstrates an efficient method for correctly assigning bond orders to small molecules extracted from the Protein Data Bank (PDB).
2018-09-24	Some Notes From the 2018 RDKit UGM	Blogger	RDKit, Conference	Highlights and key takeaways from the 2018 RDKit User Group Meeting.
2018-09-17	A Few Updates to Free-Wilson	Blogger	Free-Wilson, SAR	Discusses modern updates and computational improvements to the classic Free-Wilson SAR analysis method.
2018-09-05	Predicting Aqueous Solubility - It’s Harder Than It Looks	Blogger	Solubility, Property Prediction	Explores the complexities and challenges of accurately predicting aqueous solubility for drug-like molecules.
2018-08-20	Scaffold Hopping? It’s Complicated	Blogger	Scaffold Hopping, SAR	Discusses the theory and practical difficulties of successful scaffold hopping in medicinal chemistry.
2018-08-08	Filtering Chemical Libraries	Blogger	Library Filtering, Drug-likeness	Reviews common rules and techniques for filtering large chemical libraries to identify drug-like leads.
2018-06-08	Cheating at Word Cookies with Python	Blogger	Python, Fun	A fun diversion demonstrating how to use Python to solve word puzzles.
2018-05-30	Free Wilson Analysis	Blogger	Free-Wilson, SAR	Explains the principles and application of Free-Wilson analysis for exploring structure-activity relationships.

Share on

Twitter Facebook LinkedIn

AI in Drug Discovery - Please Stop Fishing in the Bathtub!

9 minute read

Published: February 07, 2026

The Hype Cycle and the Reality of Virtual Screening

Working on machine learning in drug discovery can be profoundly frustrating. Every week, the purveyors of social media hype herald a new “revolution” in AI for the field. As someone deeply invested in this space, I often follow the trail to these revolutionary papers, only to find researchers repeating the same fundamental mistakes that have plagued the discipline for more than a decade. Read more

Practical Cheminformatics with Marimo

7 minute read

Published: January 17, 2026

Introduction
As someone working with drug discovery data, I frequently need a view that provides an overview and then quickly switches to a detailed perspective. I’ve long been frustrated by most software tools’ inability to provide this kind of view. Most systems for processing drug discovery data adopt a spreadsheet view that places compounds in rows and assays in columns. While spreadsheet views can be powerful, they often don’t enable users to quickly switch between summary and detailed views, which are critical for understanding SAR. Read more

Can Gemini Search the ChEMBL Database?

6 minute read

Published: January 09, 2026

Introduction
I’ve been an enthusiastic user of the ChEMBL database for more than a decade. ChEMBL contains a treasure trove of information curated from medicinal chemistry journals. When I’m building an initial machine learning model or analyzing the SAR around a particular target, ChEMBL is the first place I go. I’m reasonably good at SQL, but I still spend a lot of time figuring out which data is in which tables and how to join them correctly to get the information I need. For years, I’ve longed for a more straightforward way to query ChEMBL. I recently discovered that the Gemini CLI from Google can write SQL queries for ChEMBL and extract the information I need into a CSV file. Is this my dream come true? Can I stop spending time staring at the ChEMBL schema and just pose a simple natural-language query like this?

> get the smiles,chembl_id, target_name, publication year, article doi,
and IC50 for all kinase inhibitors published after 2022
and write this into a file called kinase_inhibitors_after_2022.csv

Read more

Performing Exploratory Data Analysis on the OpenADMET ExpansionRx Blind Challenge Dataset

3 minute read

Published: November 08, 2025

When working with a new dataset, many people quickly jump into building a machine learning model. I prefer to start with exploratory data analysis (EDA) to gain a deeper understanding of the data. To address this need, I created a notebook that performs initial EDA on the OpenADMET ExpansionRx Blind Challenge Dataset. Instead of using Jupyter for this analysis, I’m using marimo, a new open source data science notebook environment that enables the creation of interactive data apps with minimal code. I think of marimo as a “better Jupyter” because it offers several features that simplify building interactive data apps, including built-in support for Altair charts, an enhanced table view, and interactive widgets. I’m working on a repository titled “Practical Cheminformatics with marimo,” which demonstrates some ways to use marimo for cheminformatics tasks. This code should be ready in a couple of weeks. Please consider this notebook a preview of what’s to come. For those interested in learning more about marimo, I recommend starting with the following resources. Read more

Pat Walters