Mitochondrial Drug Discovery Data Package with AI-Designed Molecules

10 manuscripts provide data for AI-driven drug discovery targeting mitochondrial disease. The package contains 25 characterized targets, 201 curated bioactive compounds, and 15 de novo designed candidate molecules. Julian Borges from DrugSynth AI Research deposited this preprint data in April 2026.

Use Cases

• Validate de novo designed candidate molecules using the reported 73% in silico pass rate.
• Analyze ADMET baseline profiles for 201 curated bioactive compounds.
• Train models for binding pocket characterization using the provided pathway network data.
• Benchmark AI platform performance against the included platform performance metrics.
• Identify privileged fragment scaffolds, noting the 50% novelty rate versus ChEMBL.

Strengths

• Includes 25 characterized mitochondrial targets for focused analysis.
• Provides 90 privileged fragment scaffolds, half of which are novel compared to ChEMBL.
• Contains validation results for 15 candidate molecules with a 73% pass rate.

Limitations

• Novel compound SMILES are restricted pending patent filing, limiting chemical structure access.
• Data is from preprint manuscripts and may not reflect final published findings.
• The dataset size, row count, and specific file formats are unknown.

Provenance

Source

DrugSynth AI Research

Collection Method

Integrated data from 10 manuscript datasets covering a computational drug discovery pipeline.

Freshness

Last updated in April 2026.