1Load▼
Use Load file on the 3D view (
.xyz, .feather, .json, Gaussian .log or ORCA .out), drop a file on the molecule, or paste XYZ. The bundled example is unsubstituted benzene (basic.feather). Atom numbers are 1-based (Gaussian style).Extraction runs on every molecule file in this folder:
.feather, structured .json, Gaussian .log, ORCA .out or .xyz. The 3D view is only for picking atoms.
Optional: check that picked atoms are the same element in every file. This loads the whole folder, so it only runs when you ask.
2Pick atoms▼
Picked groups table
| field | groups |
|---|
Click or type atoms
Building group
(empty)
1-indexed. One group: 1,4 · several: 1,4 4,1. Alt+Right / Alt+Left cycles the active field.
3Extract▼
Config output
▶ How to use
Extract CSV runs feature extraction and shows the table here. Use Download CSV to save it.
Model tab: pick an existing column or paste outputs, then run linear regression.
Save picks writes
Python tab = paste into a notebook as
Model tab: pick an existing column or paste outputs, then run linear regression.
Save picks writes
run_config.json if you want to rerun later:python descriptor_extractor.py --config run_config.jsonPython tab = paste into a notebook as
answers_dict.
Advanced
AWorkflow▼
Start from the closest workflow, then adjust individual engines only if needed. Default is Feather only.
1 choose workflow
2 pick atoms
3 enable engines
4 export config
5 model CSV
BThresholds▼
CFeature packages▼
Tick extra descriptor engines to include in the run (each needs its pip install).