In SupData.xlsx
, there are 11 sheets:
D_mapping
: drug index, PubChem identifier, synonyms, and IUPAC nameP_mapping
: Protein index, GeneID, and gene symbolSE_mapping
: Side effect index and nameSE_MeSH_category
: Side effect index and name, as well as MeSH category symbol and nameSE_cluster
: Side effect index and cluster nameSE-GO_adj_with_#gene
: Each row is a side effect, and each column is a GO item. The value in each cell is the number of genes identified by explainer for the side effect (row) in the GO(column).sepsis_hi
: Explanations for sepsis with high PPIUsepsis_low
: Explanations for sepsis with low PPIUcluster_enrich_mesh
: MeSH category, cluster center, P-value of over-representation of MeSH category based on Fisher's exact test, odds ratio from Fisher's exact test, cluster size, Benjamini-Hochberg adjusted P-value, and category nameCTD_permutation_test
:- side effect index and name
- permutation P-value for precision and Jaccard index
- precision
- Jaccard index
- mean precision and Jaccard index of 10,000 random explanations
- random model: nongo_strong (entire subgraph, not necessarily GO-enriched; preserve target and non-target numbers), nongo_weak (entire subgraph, not necessarily GO-enriched; not preserving target and non-target numbers), go_strong (subgraph with GO-enriched genes only, preserve target and non-target numbers), go_weak (subgraph wit GO-enriched genes only, not preserving target and non-target numbers)
- number of predicted proteins, genes in CTD associated with the side effect (disease), target genes and non-target genes in the explanation, and random explanations
- 100% * (Precision_predicted - Mean(Precision_random)) / Mean(Precision_random)
- 100% * (Jaccard_predicted - Mean(Jaccard_random)) / Mean(Jaccard_random)
cluster_enrich_GO
: Gene Ontology term, cluster center (exemplar), P-value of enrichment of GO term in cluster and odds ratio based on Fisher's exact test, cluster size and Benjamini-Hochberg adjusted P-value
Beside, the APRILE's explanations for the top 200 polypharmacy side effect predictions (ranked by predicted probability score) in training set are provided in exp_top200_train.csv
.