AspirinCode/papers-for-molecular-design-using-DL

What is this?

This GitHub project is a curated paper list — not software you run, but a reading index. It collects and organizes research papers about using generative AI and deep learning to design molecules, especially for drug discovery and molecular shape prediction (called “conformation generation”).

What does it cover?

The list is split into major themes:

• Molecular & drug design — papers on generating new molecules that could become drug candidates.
• Molecular conformation generation — predicting the 3D shape a molecule folds into, which matters for how it behaves.
• Drug-likeness metrics — sections for scoring tools like QED (a measure of how “drug-like” a molecule is) and SAscore (synthesis difficulty).
• Deep-learning methods — papers grouped by technique: RNNs, transformers, generative adversarial networks (GANs), diffusion models, reinforcement learning, and more.
• Material design — a section for AI-designed materials, though the table is currently empty (content is planned).

How is it organized?

The README has a big table-of-contents (“Menu”) linking to categories like:

• Structure-based or fragment-based molecular generation
• Text-driven generation (using text prompts)
• Multi-target and ligand-based approaches
• Quantum and omics-based methods
• Benchmark and dataset listings
• A subcollection on molecular conformation generation with its own method categories (VAE-based, GAN-based, diffusion-based, etc.)

Each category, as far as the README shows, lists papers with titles, authors, and links to the paper or journal. The project also links to a separate file (Molecular_Optimization.md) for molecule optimization work.

Notable details

• Status: The README literally says “Updating ...” — it’s an actively growing collection.
• Recent entries: Includes papers dated up to 2026 in the reviews section, showing it’s kept current.
• Related resources: The “Recommendations and References” section points to companion repositories, e.g., a protein-conformation counterpart, a list for LLMs in biomedical science, and collections on generative AI and geometry-based drug discovery.

License & contributions

• Licensed under GPL (a badge shows this).
• A badge states “contributions welcome” — the maintainers invite others to submit additions.

Who is it for?

Researchers, students, or anyone tracking the fast-moving field of AI-driven drug design. It’s a jump-off point for finding relevant papers, grouped by approach, rather than a tutorial or code base.