Home / Documentation / Quick Start

🚀 Quick Start Guide

Welcome to MOF DB! This guide will help you get started with accessing and using our MOF datasets.

Step 1: Download the Dataset

📦 Dataset Usage

  • H2MOF-ML Dataset (JSON): 🎯 Use for model training — Contains GCMC simulation values (labels) for hydrogen adsorption, enabling supervised learning.
  • hMOF-ML Dataset (JSON): 🎯 Use for prediction — Apply trained models to predict properties on 137,953 hypothetical MOF structures for large-scale screening.
💡 Workflow Tip: Train your model on H2MOF-ML, then use it to make predictions on hMOF-ML for high-throughput screening.

You can download datasets in two ways:

Option A: Web Interface

  1. Open the dataset page in your browser
  2. H2MOF-ML Dataset hMOF-ML Dataset to visit the Hugging Face dataset link.
  3. Click the download button on the page
  4. Save the JSON file to your computer

Option B: API Access

Use the following Python script to fetch the JSON directly from Hugging Face into your current working directory. Choose the appropriate script based on which dataset you need:

download_h2mof.py 
# Download H2MOF-ML Dataset
from pathlib import Path
from huggingface_hub import hf_hub_download

current_dir = Path.cwd()

file_path = hf_hub_download(
    repo_id="ytbai/H2MOF-ML",
    filename="H2MOF-ML.json",
    repo_type="dataset",
    local_dir=str(current_dir),
    local_dir_use_symlinks=False,
)

print("Saved at:", file_path)
download_hmof.py 
# Download hMOF-ML Dataset
from pathlib import Path
from huggingface_hub import hf_hub_download

current_dir = Path.cwd()

file_path = hf_hub_download(
    repo_id="ytbai/hMOF-ML",
    filename="hMOF-ML.json",
    repo_type="dataset",
    local_dir=str(current_dir),
    local_dir_use_symlinks=False,
)

print("Saved at:", file_path)

Run the corresponding script (e.g., python download_h2mof.py or python download_hmof.py) and the dataset will be saved where you execute the script.

Step 2: Load and Explore the Data

Use Python to load the JSON file and access samples with convenient dot notation, similar to the explore.py helper shown below:

explore.py 
import json
import logging
import os
from types import SimpleNamespace

logging.basicConfig(level=logging.DEBUG, format='%(asctime)s - %(levelname)s - %(message)s')
logger = logging.getLogger(__name__)
logger.disabled = True

def dict_to_namespace(d):
    """Convert a dictionary to dot-notation objects"""
    if isinstance(d, dict):
        return SimpleNamespace(**{k: dict_to_namespace(v) for k, v in d.items()})
    elif isinstance(d, list):
        return [dict_to_namespace(item) for item in d]
    return d

def load_json_samples(json_file):
    if not os.path.exists(json_file):
        raise FileNotFoundError(f"JSON file not found: {json_file}")

    with open(json_file, 'r', encoding='utf-8') as f:
        json_data = json.load(f)

    samples = {}
    for sample_name, sample_data in json_data.items():
        samples[sample_name] = dict_to_namespace(sample_data)
    return samples

json_file = "H2MOF-ML.json"
samples = load_json_samples(json_file)

sample_name = "ABEXOW"
print(samples[sample_name].lattice.abc)
print(samples[sample_name].properties.metal_composition.total_metal_atoms)
print(samples[sample_name].properties.UV_at_TPS)

# Example output
# [12.002, 11.008, 19.473]
# 7
# 23.96

💡 Tip

The loading method is the same for both datasets - simply replace the file path with the corresponding dataset filename. For detailed attribute descriptions and available properties, please refer to the Data Format documentation.

Next Steps