Notes.
Explainers, method comparisons and notes on papers and webinars. They argue and summarise rather than walk you through a run; the step-by-step material is under Tutorials. Notes that summarise someone else's talk or paper say so at the top and link the source. Entries below are in technical review and currently noindex.
Computational chemistry
ISB-N001Chem Honest QSAR Models Start With the Assay, Not the Algorithm Notes on an Optibrium webinar: curate the assay, match the method to the noise, score the test set once, plot instead of R², and respect the applicability domain. intermediate QSAR · 9 min ISB-N002Chem Chemical Databases Every ML Scientist Should Know for Drug Discovery ChEMBL, BindingDB, PubChem, PDBbind and BRENDA answer different questions. The useful choice is the table that holds the measurement you mean to model. beginner Databases · 5 min ISB-N003Chem How to Build and Validate 3D-QSAR Models: Insights From Xu et al. (2020) A 3D-QSAR model maps aligned fields onto potency. Statistics say whether it predicts; contours say where to change the molecule. Both rest on the alignment. intermediate QSAR · 5 min ISB-N004Chem How to Use DataWarrior for Drug Discovery: Key Workflows From the Villoutreix Tutorials DataWarrior is the triage step, not the model. A guide to the import-and-filter path in Bruno Villoutreix’s tutorial series, and where it sits in a project. beginner Cheminformatics · 3 min ISB-N005Chem A Practical Guide to QSAR Model Validation: Internal, Cross, and External Checks A three-tier QSAR validation framework: internal statistics, cross-validation and external checks, with the cuts papers report and where each one comes from. intermediate QSAR · 5 min ISB-N006Chem Chemical Space Is Not One Map: PCA, t-SNE, UMAP, and What They Actually Show A chemical-space plot is a chain of choices: representation, distance, projection. PCA, t-SNE and UMAP keep different structure, and a 2D map is not a similarity measure. intermediate Cheminformatics · 11 min ISB-N007Chem Bioisosteric Replacement In Silico: From Classical Rules to Data-Driven Suggestions A replacement is bioisosteric for one target and one liability, not in general. Property-space neighbours, matched pairs and SwissBioisostere give the idea list. intermediate Medicinal chemistry · 5 min ISB-N008Chem 3D-QSAR Explained: CoMFA vs CoMSIA What CoMFA and CoMSIA actually compute, how their fields differ, and what that means for alignment, PLS models and contour maps. intermediate QSAR · 5 min ISB-N009Chem BindingDB for Computational Drug Discovery: From Experimental Affinity Data to ML and Docking Benchmarks BindingDB stores measured affinities. Michael Gilson's BindingDB webinar is a guide to which rows can supervise a docking or ML job, and which lookalikes cannot. intermediate Databases · 8 min ISB-N010Chem Making AI Work in Discovery Chemistry: Precision, Trust, and Practical Value Notes from an Optibrium panel on AI in discovery chemistry: value over marginal accuracy, leakage and open benchmarks, plausible errors, and what LLMs change. beginner Machine learning · 3 min ISB-N011Chem BindingDB Meets PubChem: A Practical Workflow for Mining Bioactivity and Patent Chemistry BindingDB holds the measurement; PubChem holds the crosswalk. Christopher Southan's webinar follows a patent series from one to the other without counting deposits twice. intermediate Databases · 6 min