Analyzing results of Genomics LRB Benchmarking

Use the notebook lrb_results_analysis.ipynb to analyze results of the initial model benchmarking; slice results by distances to transcription start sites (TSS) and different annotations (e.g., promoter, enhancer, intron/exon, etc.).

Getting started

Step 1: Clone this repository:

git clone git@github.com:kuleshov-group/genomics-lrb-viztool.git
cd genomics-lrb-viztool

Step 2: Create conda env and install requirements:

conda create -n lrb_env python=3.10
conda activate lrb_env
pip install -r requirements.txt

Step 3: Download and unzip results files from the HF dataset:

wget -O results_with_annotations.zip https://huggingface.co/datasets/InstaDeepAI/genomics-long-range-benchmark/resolve/main/results_with_annotations.zip?download=true
unzip results_with_annotations.zip

Start the notebook server and use this notebook lrb_results_analysis.ipynb:

jupyter notebook

Run all cells. Use the widgets to analyze different tasks by various splits.