Create README.md

geom-drugs processed from `drugs_crude.msgpack`

- for each molecule, keep the **30 lowest-energy conformers**
- split conformers with the published `geom_permutation.npy`
https://github.com/ehoogeboom/e3_diffusion_for_molecules/blob/fce07d701a2d2340f3522df588832c2c0f7e044a/build_geom_dataset.py
https://github.com/ehoogeboom/e3_diffusion_for_molecules/raw/fce07d701a2d2340f3522df588832c2c0f7e044a/data/geom/geom_permutation.npy

- use an **80/10/10** train/validation/test split
- zero-center each conformer
- align each conformer with PCA

README.md

@@ -0,0 +1,4 @@
+---
+tags:
+- chemistry
+---