ErwanR123/alphafold_drug_discovery

This is a tutorial-grade application with zero adoption signals (0 stars, 0 forks, 6 days old, 0 velocity). It applies existing, well-established models (ESMFold, AlphaFold2) to a specific use case (SARS-CoV-2 3CLpro) with standard analysis techniques (pLDDT scoring, RMSD validation). The workflow—predict structure, score confidence, compare to experimental data—is a textbook application of published methods, not a novel contribution. No custom algorithm, no domain-specific innovation, no specialized infrastructure. This reads as a course project or personal experiment demonstrating competency with existing tools. Frontier labs (DeepMind, OpenAI, Anthropic, Google) have already published and deployed superior structure prediction models; they would not compete with this prototype but could trivially subsume its functionality into their platforms. The high frontier risk reflects that protein folding for drug discovery is a core frontier competency—labs actively build and integrate these capabilities as native features. No moat, no switching costs, no community. Easily replicated by anyone with ESMFold installed.