AI Blocks › Research a topic properly
Drug docking pose ranker
Give it a protein and a molecule. Get predicted binding poses ranked by confidence, with checks on each.
Drug docking pose ranker is a free AI skill for the job "Research a topic properly". Give the instructions to an AI assistant that supports the required files and tools, then review its output.
Free · No account · MIT licence · Made by K-Dense · Level: expert · Thinking: high
Made by K-Dense. See the original on GitHub; we copied it as it is, with its licence.
What it does
Predicted ways a small molecule could sit in a protein, ranked and labelled with how far to trust them. Give it a protein and a molecule, and you get candidate binding poses with confidence scores and checks on each. The skill runs DiffDock, a published docking model, from its official code. It sets up the exact environment the model needs, runs one pair or a batch, and can try several protein shapes. It states plainly that confidence is not a measured probability and not binding strength. Sorting a large library by score is treated as rough triage, not finding hits. You need a GPU-ready setup, and some steps are untested recipes.
You give it
- A protein structure or sequence
- A ligand, such as a SMILES string
- A GPU environment with the DiffDock setup
- Alternate protein shapes (optional)
You get
- Ranked poses with confidence scores
- Per-pose checks
- Run commands for the DiffDock setup
- Notes on what the scores can and cannot show
Use it when
- You want likely binding poses for one protein and ligand pair.
- You have a batch of molecules to dock against one target.
- You need a pose check before spending lab time.
- You want to know what the confidence score really means.
How to use it
- Add it to your AI once: download it and add the folder to Claude, ChatGPT or Codex as a skill.
- Ask for the job in your own words. Your AI opens the skill when it fits.
- It works best when your AI can run DiffDock on a computer with a strong graphics card.