Our collaborators found configuring ALLEGRO for their experiments inconvenient, and rightfully so. So I made a Claude AI assistant for it.
ALLEGRO has a big and confusing config – In an ideal world, you describe your experimental needs to the ghost in the machine, and It Just Works™
Here's how you can do exactly that:
- Get an Anthropic API key (you need $5 for this) and SAVE it (video demo below)
- Set your API key for this terminal session
export ANTHROPIC_API_KEY=sk-ant-api... - Clone this repo
- Change directory
cd Agentic-ALLEGRO - Install the dependencies
pip install --no-user anthropic allegro-bio - Download the ALLEGRO Jumpstart dataset
git clone https://github.com/AmirUCR/allegro-data.git - Run Claude x ALLEGRO
python main.py - You are now talking to Claude. Ask it to
describe ALLEGRO's parameters - Once you read what options there are, explain where your input is, and how you want your experiment set up. Here's what I asked for:
"my input is under allegro-data/data/input/example_input/ - manifest under allegro-data/data/input/two.csv - experiment is cas9 with NGG and multi = 2 - track e - no offtarget analysis, no scoring"
- You may also run a single query
python main.py --once "describe ALLEGRO's parameters" - Sit back and watch the magic happen
agentic.allegro.mov
If the embedded video above doesn't work, here's a mirror
If you run into any issues or have suggestions for Agentic ALLEGRO, please report them on our GitHub Issues tracker. It's the fastest way to get support and helps us improve Agentic ALLEGRO for everyone.
Agentic ALLEGRO has been developed and is maintained by Amirsadra Mohseni and Stefano Lonardi at the University of California, Riverside.