AI-Teachathon | Halbritter Lab

Appearance

Project Ideas ​

These are real project ideas from participants. Pick one that sounds interesting, or bring your own - the goal is to learn Git, GitHub, and AI tools by building something you care about.

An educational game about rare kidney diseases, built with web technologies and AI assistance.

KidneyQuest is an interactive game that teaches players about rare kidney diseases. It connects to the CeRKiD zebra mascot (because rare diseases are like zebras - not horses). We'll build it together during the workshop using modern web tools and AI assistance.

This is the main project everyone will work on together. Follow the hands-on guide to get started.

What you'll learn:

  • Setting up a web project with AI help
  • Using GitHub for version control
  • Building interactive features step by step
  • Debugging and iterating with AI assistance

Alternative Project Ideas ​

Looking to explore something different? These ideas come from fellow participants.

Genomic Variant Effect Prediction ​

Use Google DeepMind's AlphaGenome to predict how patient variants affect gene expression, chromatin accessibility, and histone modifications.

The right starting point depends on your question. For a visual, no-code exploration, try AlphaGenome Viewer - a web app that lets you query interval predictions, compare reference vs. alternate alleles, and score variants with point-and-click controls. For custom analyses or batch processing, the official Colab notebooks let you script against the AlphaGenome SDK directly.

AlphaGenomePythonVariant Analysis

Research Data Management Dashboard ​

Build a desktop app that handles data entry, stores everything in a local database, runs automated analysis, and displays results in visual dashboards.

Perfect if you work with structured research data and want a custom solution for your lab's workflow. For a reference implementation with a very similar stack, see VarLens - an offline-first Electron + Vue + SQLite app for clinical variant analysis. You'll learn how to build cross-platform desktop apps, work with databases, and create interactive visualizations.

ElectronVue/ViteSQLiteDesktop Tools

AI-Assisted Academic Writing Pipeline ​

Combine Zotero (for managing references), QuillBot (for paraphrasing), and GPTZero (for checking AI detection) into a streamlined research writing workflow.

Learn how to chain different AI tools together and build a personal writing assistant. This project focuses on the practical workflow of writing papers: organizing references, improving clarity without plagiarism, and ensuring your writing sounds human.

ZoteroQuillBotGPTZeroAcademic Writing

AI-Powered RNAseq Enrichment Summarizer ​

Feed enrichment analysis results to an LLM via API, get plain-language summaries of what the results mean, then validate those summaries using an LLM-as-judge approach.

Covers the full LLM workflow: designing prompts, handling structured data, avoiding hallucinations, and validating outputs. You'll learn how to use AI APIs programmatically, not just through chat interfaces.

PythonOpenAI/Anthropic APIPrompt Engineering

What Could AI Help With? ​

Don't have a specific project yet? Here are common research pain points where AI tools can help. Pick one that matches your daily frustrations and explore it during the hands-on session.

Common Research Pain Points ​

Literature Review ​

Ask AI to read papers, extract key findings, and compare methodologies across studies. Great for getting up to speed on a new topic.

Data Cleaning ​

Transform messy datasets into tidy formats, handle missing values, or convert between file types. AI can write scripts to automate tedious data prep.

Manuscript Writing ​

Generate first drafts of methods sections, improve clarity, or adapt tone for different audiences. Overcome writer's block and polish rough drafts.

Debugging Code ​

Paste error messages and get explanations, suggested fixes, and prevention strategies. AI excels at pattern matching in error logs.

Boilerplate Code ​

Create starter templates for common tasks like data visualization, statistical tests, or file I/O. Save time on repetitive patterns.

Data Visualization ​

Describe what you want to show, and AI can suggest plot types and generate the code. Useful for exploratory analysis or publication figures.

Project-Specific Resources ​

Contribute Your Ideas ​

Have a project idea? Edit this page and add it - that's what contributing on GitHub looks like.

Next step: Check the resources page for curated links and further learning.