extract_feature_vectors¶

Batch-extract structural feature vectors for annotated LeuT transporter PDB files.

Synopsis¶

python scripts/extract_feature_vectors.py [OPTIONS]

Description¶

Reads the LeuT annotations CSV and PDB files from an input directory, runs confostate.features.extract_features_batch(), and writes a CSV of feature vectors suitable for ML training (Person 3).

Structures must be downloaded first with scripts/download_structures.py.

Options¶

Flag	Default	Description
`--annotations`	`data/annotations/leu_t_transporters.csv`	Annotations CSV path
`--input-dir`	`input`	Directory containing `.pdb` files
`--output`	`data/features/leu_t_feature_vectors.csv`	Output feature CSV path

Examples¶

cp data/annotations/leu_t_transporters.csv.example \
   data/annotations/leu_t_transporters.csv

python scripts/download_structures.py \
    --codes-file data/protein_families/LeuT_transporters.txt \
    --output-dir input

python scripts/extract_feature_vectors.py

Output¶

Writes a CSV to data/features/leu_t_feature_vectors.csv with one row per structure. Columns include pdb_id, conformation, cavity/domain/RMSD/orientation features, and file_path.