Integrating DBRetina with an LLM-Powered Chatbot for Genomic Data Exploration

[MoHelmy]

Introduction to DBRetina

DBRetina is a high-performance bioinformatics tool with an efficient linear algorithm for calculating the pairwise distance among large collections of gene sets. This algorithm enables easy construction of a comprehensive pairwise molecular similarity network within and across several molecular databases. To enable efficient search and visualization of this huge similarity network, DBRetina can transform the final output into a format compatible with the Neo4j graph databases.

Challenge:

While DBRetina bridges genomic analytics and graph databases, querying Neo4j requires Cypher query language expertise, limiting accessibility for non-technical researchers.

Goal and Aims

To develop an LLM-Driven chatbot that translates natural language questions into Cypher queries, enabling intuitive interaction with DBRetina-generated Neo4j graphs.
This chatbot aims to:
Increase accessibility: Enables non-technical users to query complex genomic networks.
Improve efficiency: Reduces query-writing time by ~70% (based on LLM benchmarks ).
Scalability: Adapts to evolving graph schemas and supports multi-database integration.

Difficulty Level: Medium/Hard

Size and Length of Project

• medium: 175 hours
• 12 -16 weeks

Skills

Essential skills: LLM fine tuning, Experience with Graph databases, HTML, CSS, JS
Nice to have skills: C++

Public Repository

DBRetina Documentation
Neo4j Cypher Manual
LLM Fine-Tuning for KBQA

Potential Mentors

Mohamed Helmy
Tamer Mansour

[harshagr70]

Hi @MoHelmy , I was going through the projects listed by you, and while I believe I have the skills to contribute to either of them, this particular project excites me the most. Having previously worked on projects involving fine-tuning LLMs and exploring graph-based tools like Cytoscape, I feel this aligns well with my interests and experience.

Additionally, I am currently brushing up on my skills through active contributions to open-source projects, and also contributing actively in machine learning domain . I am very keen to contribute to this project and am confident in my ability to learn and deliver effectively.

Looking forward to hearing your thoughts.

[Jitmandal051004]

Hi @MoHelmy , I’d like to contribute to this issue. I have experience in building full-stack websites and have previously worked on developing RAG pipeline for local LLM as part of my robotics project. Looking forward to your response

[MoHelmy]

@harshagr70 Please write to us on the provided email addresses.

[MoHelmy]

@Jitmandal051004 Please write to us on the provided email addresses.

[Jitmandal051004]

Hi @MoHelmy , Just following up on my email from February 3rd. Could you please take a look and reply when you get a chance?

[MoHelmy]

@Jitmandal051004 received and will write to you soon.