Repository with the data and code to reproduce the evaluation of computational methods to predict functional intronic variation described in the paper: https://academic.oup.com/gigascience/article/doi/10.1093/gigascience/giad085/7329463
data folder contains all the datasets in the hg19 version used for all comparisons. VCF files are annotated with predictions from the tools included in the study.
data_hg38 folder contains the main datasets in the hg38 version (via liftover).
intronic_benchmark.ipynb contains the code to reproduce most of the analysis. Main results were generated with VETA, a tool to evaluate the performance of variant effect prediction tools in multiple contexts.
scripts folder contains the post-processing scripts used to create main figures. These scripts take as input results generated by the intronic_benchmark.ipynb notebook.
To fully reproduce the results we recommend using Docker. In the command line simply run:
git clone https://github.com/PedroBarbosa/DeepIntronic_Benchmark
cd DeepIntronic_Benchmark
./build_docker.sh
./run_docker.sh
This will create a Docker image with all software (and exact versions) that will run the intronic_benchmark.ipynb jupyter notebook. All results will be saved in the out folder.