ALCHIMIA: Generated Molecules for CB2R and S1R Targets with QED and SA Scores

1.8 MB of CSV data contains molecules generated by the ALCHIMIA framework for two pharmacologically relevant targets. The dataset, created by Domenico Alberga and last updated in April 2026, likely contains results from applying a hybrid reinforcement learning and genetic algorithm method. It was designed to optimize target activity, synthetic accessibility, and drug-likeness in molecular de novo design.

Use Cases

• Benchmarking generative molecular design algorithms based on reported QED and SA scores.
• Training or validating reinforcement learning policies for chemical transformation sequences.
• Analyzing chemical diversity in generated molecular populations based on the genetic algorithm component.
• Exploring molecular structures for the CB2R and S1R targets mentioned in the description.

Strengths

• Dataset is 1.8 MB, making it a small and easily downloadable file.
• Framework is based on a vocabulary of 33 medicinal chemistry-inspired molecular transformations.
• Results are validated against two specific pharmacological targets: human Cannabinoid Receptor 2 (CB2R) and Sigma Receptor 1 (S1R).
• Source code for the ALCHIMIA framework is publicly available on GitHub, aiding reproducibility.

Limitations

• Row count is unknown, which may limit suitability assessment.
• Column-level documentation is absent; field semantics must be inferred after download.
• Description metadata is limited; actual data quality requires manual inspection after download.

Provenance

Source

Domenico Alberga via figshare

Collection Method

Generated by the ALCHIMIA computational framework combining reinforcement learning and a genetic algorithm.

Freshness

Last updated 2026-04-16 07:29:45; freshness should be verified.