Skip to content

PACKAGE_STRUCTURE.md

Latest commit

 

History

History
93 lines (71 loc) · 2.73 KB

PACKAGE_STRUCTURE.md

File metadata and controls

93 lines (71 loc) · 2.73 KB
Raw

Package Structure

Your ConfoState project now follows a Python package workflow:

Directory Layout

ConfoState/
├── confostate/                 # Main Python package
│   ├── __init__.py            # Package initialization
│   └── data/                  # Data handling subpackage
│       ├── __init__.py
│       └── loader.py          # CSV loading utilities
├── data/
│   └── annotations/           # Annotated datasets
│       └── leu_t_transporters.csv  # 25 LeuT transporters with states
├── input/                      # Input directory for PDB files (empty initially)
├── examples/
│   └── load_data.py           # Example: load and inspect data
├── scripts/
│   └── download_structures.py  # Download PDB files from RCSB
├── pyproject.toml             # Package configuration
└── README.md                  # Main documentation

Quick Start

1. Install the package (already done)

pip install -e .

2. Download structures to the input directory

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

3. Load and inspect the data

python examples/load_data.py

4. Use the package in your code

from confostate.data.loader import load_annotations, load_from_input_dir

# Load annotations
df = load_annotations("data/annotations/leu_t_transporters.csv")
print(df.head())

# Load structures from input directory
df_structures = load_from_input_dir("input")
print(f"Found {len(df_structures)} structures")

What's Included

Annotated Dataset: LeuT Transporters (25 structures)

The CSV file contains:

  • pdb_id: PDB identifier
  • family: Protein family
  • conformation: Conformational state (OF_open, IF_open, Occluded, Intermediate)
  • reference: Literature reference
  • experimental_method: X-ray or Cryo-EM
  • resolution_angstrom: Crystal/EM resolution
  • year: Publication year

States are based on the alternating access model:

  • OF_open: Outward-facing, open to substrate
  • IF_open: Inward-facing, open to substrate
  • Occluded: Substrate bound, closed on both sides
  • Intermediate: Transitional conformation

Data Loader (confostate.data.loader)

Functions:

  • load_annotations(csv_path, family=None) — Load CSV with optional filtering
  • load_from_input_dir(input_dir) — Scan input directory for PDB files

Next Steps

  1. Download all PDB structures to input/ directory
  2. Add feature extraction modules to confostate/features/
  3. Build ML classifier in confostate/models/
  4. Add explainability layer in confostate/explain/
  5. Create CLI in confostate/cli.py