This repository is for our paper: Causal relationships between diseases mined from the literature improve the use of polygenic risk scores
The paper discusses the creation of a Directed Acyclic Graph (DAG) for causal relations between diseases mapped to ICD-10-CM identifiers and its use to improve polygenic risk scores. An interactive visualization of the DAG can be found here.
The used data is available as follows:
- The Disease Ontology version we used is available through this link
- The main DAG with various metrics included can be found in:
- data/DDC_DAG.tsv
- The full cyclic graph
- data/DDC_full_graph.tsv
- The used dictionary of ICD10 names can be found in:
- data/disease_dict.json
- The expert curated data that was used for evalution can be found in:
- data/curated_positive_relations.tsv for curated positive relations
- data/curated_negative_relations.tsv for curated negative relations
- The autmatically generated Polygenic Risk Scores (PRSs) can be downloaded via this link