README.md

---
license: other
license_name: genereviews
license_link: https://www.ncbi.nlm.nih.gov/books/NBK138602/
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
dataset_info:
  features:
  - name: id
    dtype: string
  - name: ch_id
    dtype: string
  - name: keywords
    list: string
  - name: title
    dtype: string
  - name: authors
    dtype: string
  - name: abstract
    dtype: string
  - name: content
    dtype: string
  - name: references
    list: string
  - name: created_date
    dtype: string
  - name: updated_date
    dtype: string
  - name: revised_date
    dtype: string
  - name: journal
    dtype: string
  - name: source_url
    dtype: string
  - name: publication_types
    list: string
  splits:
  - name: train
    num_bytes: 57554171
    num_examples: 929
  download_size: 15286246
  dataset_size: 57554171
language:
- en
tags:
- medical
- gene
- reviews
- medicine
pretty_name: 'GeneReviews '
size_categories:
- n<1K
task_categories:
- text-generation
---

# GeneReviews Dataset Extraction

This project extracts text and metadata from GeneReviews® chapters downloaded from NCBI Bookshelf and creates a structured dataset in Hugging Face format.

## Overview

GeneReviews® is an international point-of-care resource for clinicians, providing clinically relevant and medically actionable information for inherited conditions. This project processes the XML files from the GeneReviews database and creates a structured dataset suitable for machine learning and research applications.


### 📈 **Dataset Statistics:**

- **Total Records**: 929 GeneReviews chapters
- **Average Abstract Length**: 899.6 characters
- **Average Content Length**: 56,377.9 characters
- **Total References**: 13,683 references across all chapters
- **Average References per Chapter**: 14.7
- **Chapters with >100 references**: 12 chapters
- **Total Keywords**: 9,616
- **Unique Keywords**: 6,824

## Source Information

- **Source**: [GeneReviews® on NCBI Bookshelf](https://www.ncbi.nlm.nih.gov/books/NBK1116/)
- **Publisher**: University of Washington, Seattle
- **ISSN**: 2372-0697
- **Content Type**: Clinical reviews of genetic conditions
- **License**: Open access for noncommercial research purposes

## Dataset Structure

Each record in the dataset contains the following fields:

| Field | Type | Description |
|-------|------|-------------|
| `id` | string | Unique chapter identifier |
| `ch_id` | string | Chapter ID (as you renamed it) |
| `title` | string | Chapter title |
| `authors` | string | Comma-separated author names |
| `journal` | string | "GeneReviews®" |
| **`abstract`** | **string** | **Chapter abstract/summary only** |
| **`content`** | **string** | **Chapter body content only (excluding abstract)** |
| **`references`** | **array** | **Array of reference citations** |
| `keywords` | array | Keywords and terms |
| `source_url` | string | Link to GeneReviews resource |
| `publication_types` | array | ["Review", "Clinical Review"] |
| `created_date` | string | Creation date |
| `updated_date` | string | Last update date |
| `revised_date` | string | Revision date |


## Files

- `extract_genereviews.py`: Main extraction script
- `load_genereviews_dataset.py`: Script to load and demonstrate the dataset
- `requirements.txt`: Python dependencies
- `genereviews_dataset/`: Hugging Face dataset directory
- `genereviews_dataset.json`: JSON version of the dataset

## Installation

1. Install the required dependencies:
```bash
pip install -r requirements.txt
```

## Usage

```python
from datasets import load_from_disk

# Load the dataset
dataset = load_from_disk("genereviews_dataset")

# Access by chapter
record = dataset[0]
chapter_id = record['ch_id']

# Access separated content
abstract = record['abstract']      # Only the abstract
content = record['content']        # Only the body content
references = record['references']  # Array of reference citations
```

### Search for Specific Conditions

```python
# Search for cystic fibrosis
cf_records = dataset.filter(lambda x: "cystic fibrosis" in x['title'].lower())

# Search for cancer-related content
cancer_records = dataset.filter(lambda x: "cancer" in x['content'].lower())
```

### Analyze Publication Dates

```python
# Find recently updated chapters
recent_updates = dataset.filter(lambda x: "2024" in x['updated_date'])
```

### Extract Keywords

```python
# Get all unique keywords
all_keywords = set()
for record in dataset:
    all_keywords.update(record['keywords'])
```

## Citation

When using this dataset, please cite:

```
GeneReviews® [Internet]. Seattle (WA): University of Washington, Seattle; 1993-2025.
Available from: https://www.ncbi.nlm.nih.gov/books/NBK1116/
```

## License

This dataset is derived from GeneReviews®, which is owned by the University of Washington. 
Permission is granted to reproduce, distribute, and translate copies of content materials 
for noncommercial research purposes only, provided that proper attribution is given.