Vol. 1 · No. 1 · October 2026 Last full re-run 23 Oct 2026 · versions pinned
InSilicoBench
ISB-N004 Computational chemistry Cheminformatics · beginner

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.

Author
Published · updated
Reading time 3 min

First published on yboulaamane.github.io on ; revised for InSilicoBench on . Status: in technical review.

DataWarrior is the triage step, not the model. Bruno Villoutreix’s nine-part DataWarrior tutorial series on his AI-Biotech-Studio channel is the guided tour this note follows [2]. Part 1 is a filter: a name pattern, then a substructure, then a row list you can export.

The practice set is the Wikipedia molecule import inside the program, on the order of 22,000 structures, depending on the version. No separate download URL is given here.

Filter twice, then exportFour steps joined by arrows, from Part 1 of the Villoutreix series. 1 Wikipedia set, imported inside the app. 2 Name filter, a suffix regular expression. 3 Substructure filter, a drawn fragment. 4 Row list, exported with SMILES.01Wikipediaimported insidethe app02Namesuffix regexe.g. .*sartan03Substructuredrawn fragment,hits in red04Row listSMILES columnout as .txtA row has to pass both filters before it reaches the list.
Filter twice, then exportFour steps joined by arrows, from Part 1 of the Villoutreix series. 1 Wikipedia set, imported inside the app. 2 Name filter, a suffix regular expression. 3 Substructure filter, a drawn fragment. 4 Row list, exported with SMILES.01Wikipediaimported inside the app02Namesuffix regex e.g. .*sartan03Substructuredrawn fragment, hits in red04Row listSMILES column out as .txt
Figure 1.The name rule and the substructure rule both have to pass. Matches are highlighted in red. The export in the tutorial is a tab-delimited file, after SMILES are added and the columns you do not want are dropped. cancer.txt is the example name for one such list.Schematic, not data.

What the program is

DataWarrior is a free, open-source, chemistry-aware table. Academic and commercial use are both covered. It filters in real time on text, numbers, and structure. It draws 2D and 3D plots. It calculates properties, enumerates combinatorial libraries, runs PCA and t-SNE, and can dock into a pocket. Formats are SMILES, SDF (v2 and v3), CSV, tab-delimited text, and its own .dwar file format.

The program was released for public use by its developers, and its design is described in the 2015 paper by Sander and colleagues [1]. The problem the paper sets out is a medicinal-chemistry one: a project accumulates thousands of structures and many assays, and someone has to see which molecular families are present, which motifs track a measured property, and which small structural changes cause large property changes. The paper also presents an unsupervised 2D scaling method that DataWarrior uses to show chemical or pharmacophore space, activity landscapes, and activity cliffs [1]. The source code is public on GitHub, and installers are on openmolecules.org.

A lot of the interface is a right-click, a sub-window, or a pop-up. The useful mental model is a spreadsheet that knows what a molecule is.

Villoutreix’s series is about 85 minutes across 9 videos. Part 1 is the import-and-filter path summarized here. The videos are where the clicks are. This page does not restage them as screenshots.

The Part 1 filter

Import the Wikipedia structures from inside DataWarrior. Then:

  1. Filter names with a regular expression. The examples are .*ib for kinase inhibitors, .*sartan for angiotensin receptor blockers, and .*azole for antifungals. Applied to the Wikipedia set, each suffix is a much smaller table.
  2. Draw a fragment in the structure filter. The example is a piperazine or piperidine. Rows have to match the name rule and contain the fragment. The hit atoms are shown in red.
  3. Make a row list. The example list is named cancer. Add a SMILES column. Drop columns you do not need, such as formula, molecular weight, and the structure depiction. Save tab-delimited text, cancer.txt.

Repeat with another suffix, such as .*azole, when you want a second set.

Where it sits in a project

Bring your own SDF or SMILES in, and merge assay or ADMET columns, once the Wikipedia exercise is familiar. The same two filters apply: a name or series pattern, and a scaffold. Plots are for activity, scatter, PCA or t-SNE, and the clusters you would otherwise miss in a table. A 2D map in DataWarrior carries the same caveat as any other projection: which structure it keeps depends on the descriptor and the method, so a cluster is a reason to look, not a measurement. The export is the clean SMILES-plus-metadata file that docking, QSAR, molecular dynamics, or a library design step can read.

Install DataWarrior from the official site, then follow the steps above on the Wikipedia set before you trust the same clicks on a project table. The program is the interactive layer in front of the modeling. It is not the model.

Questions to settle before the export

The Part 1 filter is quick, and that is the risk. Before a filtered list goes to docking or a model, it is worth writing down:

  • which version of the Wikipedia set, or which project file, the rows came from;
  • the exact regular expression and the exact drawn fragment, because both decide what was left out;
  • whether stereochemistry and salts in the SMILES column are what the next program expects.

For a bioactivity table rather than a name list, the curation steps are in the ChEMBL curation tutorial.

References

  1. Sander T, Freyss J, von Korff M, Rufener C. DataWarrior: an open-source program for chemistry aware data visualization and analysis. J Chem Inf Model 2015;55(2):460–473. doi:10.1021/ci500588j
  2. Villoutreix B. Introduction to DataWarrior (part 1), and the following eight parts. AI-Biotech-Studio, YouTube. youtube.com/watch?v=HLMko3whhAI

About the author

Yassir Boulaamane

Yassir Boulaamane, Postdoctoral researcher, computational drug discovery

Postdoctoral researcher at the Universitat Autònoma de Barcelona (InSiliChem), with a PhD in computational drug discovery. Works on molecular docking, molecular dynamics, free energy calculations and machine-learning QSAR.

Found an error or a command that no longer works? Tell me and I will correct it. See the editorial policy for how corrections are handled.

Need this workflow run on your own system? See services.

ISB-N006Chemical Space Is Not One Map: PCA, t-SNE, UMAP, and What They Actually ShowISB-0006How to Curate and Classify Bioactivity Data From ChEMBLISB-N002Chemical Databases Every ML Scientist Should Know for Drug Discovery Learn CADDModule 5: Cheminformatics & Molecular Representations