This repository contains the scripts used for the paper 'Deep learning-based prediction of tissue-specific splice sites in the human neural retina' by Riepe et al. (2024).
In Training data we describe how we obtained the retina dataset included in datasets.
- Figures and tables contains all scripts used to create the figures and tables for the manuscript
- Models contains all models that we trained. For each hyperparameter setting, five separate models were trained.
- Output test contains the test output for each model. Each model was evaluated using the retina, GTEx, and GENCODE (canonical) dataset.
- Output_train contains the training output for each model.
- Predictions contains the predictions for the retina-enriched exons, control exons, and variants with a retina-specific splicing defect.
- Reference data contains all reference files used for validation of the models. This includes the retina-enriched exons and control exons.
- Scripts contains all scripts used for training and testing the model.
- Validation scripts contains the scripts used for validating the models on the exons and variants.
- Variants contains the vcf file of the variants with a retina-specific splicing defect.