ISCB-Africa ASBCB 2025 Tutorial IP5: Building Agentic workflows in bioinformatics

This repo contains materials for the tutorial session - Tutorial IP5: Building agentic workflows for bioinformatics.
https://www.iscb.org/africa2025/home

Tutorial description

Agentic workflow is a process of interacting with Large Language Models (LLMs) to complete complex tasks - allowing practitioners to build pipelines that integrate data retrieval, reasoning, and execution steps. This tutorial guides participants through the conceptual and practical foundations of setting up their own agentic workflows.

Dataset

Dataset is taken from https://github.com/GoekeLab/bioinformatics-workflows which contains RNA-Seq reads and a transcriptome reference file.

Requirements

Either Docker or Conda/Mamba environment. You don't need both.

1. The tutorial can be run inside the Docker container. The easiest way to make sure it will work on your system is to have Docker setup on your machine.
2. Setup a Conda/Mamba environment. If you haven't done this you can follow the instruction and use miniforge to install it https://github.com/conda-forge/miniforge

Clone this repository

git clone https://github.com/dionizijefa/iscb-bioinformatics-agents.git
cd iscb-bioinformatics-agents

Conda/Mamba instruction

mamba env create -f environment.yml
mamba activate bio-agent

Docker instructions

If running from command line

1. Build the image

docker build -t bio-agent .

2. Run the directory and mount a drive (Unix)

docker run -it --volume $(pwd):/app bio-agent

Windows a) Powershell

docker run -it --volume ${PWD}:/app bio-agent

b) CMD

docker run -it --volume "%cd%":/app bio-agent

3. Don't forget to activate the micromamba environment

micromamba activate bio-agent

Running the tutorial code

The default setup requires you to use an Azure API key. To do that create a .key file in the root directory and paste the given key inside of it. After that you can create a results folder if you wish

mkdir results

Then you just run the agent with

python agent.py