NetInfer

NetInfer predicts drug–target and drug–pathway relationships, including target proteins, microRNAs, therapeutic and adverse effects, to support pathway-based drug repurposing and systems pharmacology analyses.

Key Features:

• Target Prediction: Predicts potential target proteins and microRNAs associated with specific drugs.
• Therapeutic and Adverse Effect Prediction: Predicts therapeutic effects using Anatomical Therapeutic Chemical (ATC) classification codes and adverse drug events based on models constructed from systematic evaluations in prior studies.
• Drug Repositioning: Validates predicted therapeutic effects against existing literature and has been demonstrated in a cardiovascular disease case study.
• Pathway-Based Drug Repurposing (DPNetinfer): The DPNetinfer module predicts potential drug–pathway associations using substructure–drug–pathway networks to enable large-scale pathway-based repurposing.
• Performance Metrics: DPNetinfer achieved an area under the curve (AUC) of 0.9358 in a pan-cancer network and showed generalization on external datasets with AUCs of 0.8519 and 0.7494.
• Case Studies: Identified unexpected anticancer activities of non-oncology drugs targeting the PI3K-Akt pathway and accounts for tumor heterogeneity by constructing primary site–based models across different drug–pathway networks.

Scientific Applications:

• Systems Pharmacology: Enables analysis of drug interactions with proteins, microRNAs, and pathways for systems-level pharmacological investigations.
• Drug Repurposing: Supports identification and validation of new therapeutic uses for existing drugs, including pathway-based repurposing via DPNetinfer.
• Oncology and Cancer Therapeutics Discovery: Facilitates discovery of anticancer activities in non-oncology drugs, analysis of the PI3K-Akt pathway, and modeling that accounts for tumor heterogeneity by primary site.

Methodology:

Predicts drug targets and microRNAs; predicts therapeutic and adverse effects using ATC classification codes with models constructed from systematic evaluations; DPNetinfer predicts drug–pathway associations using substructure–drug–pathway networks and constructs primary site–based models across drug–pathway networks.