# AItrika

![AItrika](images/logo.png)

[![License](https://img.shields.io/badge/License-Apache%202.0-orange.svg)](https://opensource.org/licenses/Apache-2.0)

![GitHub forks](https://img.shields.io/github/forks/dSupertramp/AItrika)

![GitHub commit activity (branch)](https://img.shields.io/github/commit-activity/t/dSupertramp/AItrika/main)

![GitHub last commit (branch)](https://img.shields.io/github/last-commit/dSupertramp/AItrika/main)

![Static Badge](https://img.shields.io/badge/medical-content?logo=syringe&logoColor=cyan&color=cyan)

Enhance your knowledge in medical research.

AItrika (formerly **PubGPT**) is a tool that can extract lots of relevant informations inside medical papers in an easy way:

- Abstract

- Full text (when available)

- Genes

- Diseases

- Mutations

- Associations between genes and diseases

- MeSH terms

- Other terms

- Results

- Bibliography

And so on!

## 🚀 Run the demo app

You can try AItrika with the Streamlit app by running:

```

streamlit run app.py

```

Or you can use it a script by running:

```

python main.py

```

## 📦 Install

To install everything, you need `uv`.

First of all, install `uv` with the command:

```

python main.py

```

After that, create a virtual environment with the command:

```

uv venv venv_name

```

Activate the virtual env:

```

source venv_name/bin/activate

```

And install dependencies:

```

uv pip install -r requirements.in

```

## 🔑 Set LLM API Keys

In order to set API keys, insert your keys into the `env.example` file and rename it to `.env`.

## 🔍 Usage

You can easily get informations of a paper by passing a PubMed ID:

```python

from aitrika.engine.aitrika import OnlineAItrika

aitrika_engine = OnlineAItrika(pubmed_id=pubmed_id)

title = aitrika_engine.get_title()

print(title)

```

Or you can parse a local pdf:

```python

from aitrika.engine.aitrika import LocalAItrika

aitrika_engine = LocalAItrika(pdf_path = pdf_path)

title = aitrika_engine.get_title()

print(title)

```

```

Breast cancer genes: beyond BRCA1 and BRCA2.

```

You can get other informations, like the associations between genes and diseases:

```python

associations = aitrika_engine.get_associations()

```

```

[

  {

    "gene": "BRIP1",

    "disease": "Breast Neoplasms"

  },

  {

    "gene": "PTEN",

    "disease": "Breast Neoplasms"

  },

  {

    "gene": "CHEK2",

    "disease": "Breast Neoplasms"

  },

]

...

```

Or you can get a nice formatted DataFrame:

```python

associations = aitrika_engine.associations(dataframe = True)

```

```

      gene                          disease

0    BRIP1                 Breast Neoplasms

1     PTEN                 Breast Neoplasms

2    CHEK2                 Breast Neoplasms

...

```

With the power of RAG, you can query your document:

```python

## Prepare the documents

documents = generate_documents(content=abstract)

## Set the LLM

llm = GroqLLM(documents=documents, api_key=os.getenv("GROQ_API_KEY"))

## Query your document

query = "Is BRCA1 associated with breast cancer?"

print(llm.query(query=query))

```

```

The provided text suggests that BRCA1 is associated with breast cancer, as it is listed among the high-penetrance genes identified in family linkage studies as responsible for inherited syndromes of breast cancer.

```

Or you can extract other informations:

```python

results = engine.extract_results(llm=llm)

print(results)

```

```

** RESULTS **

- High-penetrance genes - BRCA1, BRCA2, PTEN, TP53 - responsible for inherited syndromes

- Moderate-penetrance genes - CHEK2, ATM, BRIP1, PALB2, RAD51C - associated with moderate BC risk

- Low-penetrance alleles - common alleles - associated with slightly increased or decreased risk of BC

- Current clinical practice - high-penetrance genes - widely used

- Future prospect - all familial breast cancer genes - to be included in genetic test

- Research need - clinical management - of moderate and low-risk variants

```

## 🚀 Run the API

To run the AItrika API, follow these steps:

1. Ensure you have set up your environment and installed all dependencies as described in the Installation section.

2. Run the API server using the following command:

```bash

python api.py

```

The API will start running on http://0.0.0.0:8000. You can now make requests to the various endpoints:

- /associations: Get associations from a PubMed article

- /abstract: Get abstract of a PubMed article

- /query: Query a PubMed article

- /results: Get results from a PubMed article

- /participants: Get number of participants from a PubMed article

- /outcomes: Get outcomes from a PubMed article

You can use tools like curl, Postman, or any HTTP client to interact with the API. For example:

```bash

curl -X POST "http://localhost:8000/abstract" -H "Content-Type: application/json" -d '{"pubmed_id": 12345678}'

```

The API documentation is automatically generated and saved to <code>docs/api-reference/openapi.json</code>.

You can use this file with tools like Swagger UI for a more interactive API exploration experience.

## Support the Project

If you find this project useful, please consider supporting it:

- 🌟 Star the project on GitHub

- 🐛 Report bugs or suggest new features

- 🤝 Contribute with pull requests

- ☕️ [Buy me a coffee](https://www.buymeacoffee.com/dsupertramp) or consider a sponsor.

### Commercial / Business use

If you're using this project in a business or commercial context, please [contact me](salvatoredanilopalumbo@gmail.com).

I'm available for consulting, custom development, or commercial licensing.

Your support helps keep this project active and continuously improving. Thank you!

## License

AItrika is licensed under the Apache 2.0 License. See the LICENSE file for more details.

## Star History

[![Star History Chart](https://api.star-history.com/svg?repos=dSupertramp/AItrika&type=Date)](https://star-history.com/#dSupertramp/AItrika&Date)