Repurposing Approved Drugs as Specific Cytokine Inhibitors
Case Study 05
Literature mining, machine learning, and large-scale virtual screening were combined to prioritize known and potential inhibitors of a target cytokine.
Challenge
The client wanted CPI to identify specific inhibitors of a cytokine by combining literature mining with CPI's computational and machine-learning approaches to drug repurposing.
What We Did
-
1
Mapped evidence across research databases
Searched PubChem, PubMed, ChEMBL, BindingDB, LINCS, and other sources for molecules reported to directly or indirectly decrease the cytokine.
-
2
Prioritized known inhibitors and additional candidates
Identified FDA-approved drugs reported to inhibit the cytokine, along with hundreds of BindingDB molecules, compounds from a large PubChem HTS screen, and natural products from an NIH dataset.
-
3
Built models and screened approved drugs
Used the assembled datasets to generate machine learning models and score more than 2,000 FDA-approved drugs. Several proposed candidates were also identified in the literature as cytokine inhibitors.
-
4
Expanded screening to commercial molecules
Used virtual screening to score more than 6 million commercial molecules and identify additional potential inhibitors.