2.0
Failure to demonstrate efficacy and safety issues are important reasons that drugs do not reach the market. An incomplete understanding of how drugs exert their effects hinders regulatory and pharmaceutical industry projections of a drug's benefits and risks. Signaling pathways mediate drug response and while many signaling molecules have been characterized for their contribution to disease or their role in drug side effects, our knowledge of these pathways is incomplete. To better understand all signaling molecules involved in drug response and the phenotype associations of these molecules, we created a novel method, PathFX, to identify these pathways and predict drug-related phenotypes. We benchmarked PathFX by predicting drugs' marketed disease indications and reported a sensitivity of 41%. We then used PathFX predictions to strengthen signals for drug-adverse event pairs occurring in the FDA Adverse Event Reporting System (FAERS) and also predicted opportunities for drug repurposing for new diseases based on inter-action paths that associated a marketed drug to that disease. By discovering molecular interaction pathways, PathFX improved our understanding of drug associations to safety and efficacy phenotypes. The algorithm provides a new means to improve regulatory and therapeutic development decisions.
In the zipped folder you receive, you will find the following files:
One or more files with the ending neighborhood.txt:
These are the protein neighborhoods for the individual drug targets.
One or more files with the ending specific_neighborhood.txt:
These are the protein neighborhoods after controlling for study bias in the interaction network. For further information, please see the website.
The file ending with merged_neighborhood.txt:
This is the full drug network after controlling for study bias and considering interactions from all drug targets. The edge scores between proteins and gene variants reflect the amount and quality of evidence supporting the interaction. This edge score is explained in [1] Wilson et al, 2018
The file ending with merged_neighborhood__assoc_table.txt:
This is a table of phenotypes associated with the network.
The file ending with merged_neighborhood__assoc_database_sources.txt:
This file lists where all gene-to-phenotype associations originate.
The *.pkl files, including: lin_pandas_matrix.pkl, disease_clusters_lin_1.7.pkl, merged_neighborhood__cui_list_.pkl:
These are intermediate files from the phenotype clustering phase of the algorithm.
The cluster_membership_*.txt files:
These are tables of phenotype clusters and the phenotypes assigned to each cluster.
Two .png figures showing the results of the phenotype clustering: one file ending with labeledClusters_dendogram_full_1.7.png where the dendrogram labels show the top words associated with a particular cluster and one file ending with unlabeled_dendogram_full_1.7.png where the dendrogram labels show the number of phenotypes collapsed into a cluster or in the case of single-phenotypes clusters, the label shows the individual concept unique identifier (CUI).
A file ending with merged_neighborhood__withDrugTargsAndPhens.txt:
This is the complete protein network with phenotype interactions that is used for creating Cytoscape visualizations. The edge scores in this file are set to 1.0 for all drug-to-protein interactions and gene-to-phenotype interactions. The edge scores between proteins and gene variants reflect the amount and quality of evidence supporting the interaction. This edge score is explained in [1] Wilson et al, 2018.
A file ending with network_nodeType.txt:
This file specifies the entity type in the network file. It is also used with Cytoscape for visualization of the network.
An optional file ending with graph.json:
This file contains CytoscapeJS network visualization information created by using the Visualize Network function in this web application.
A README.txt file explaining the data in the folder and containing details of the analysis and PathFX version information (for reproducibility).