Table 1 Drug repurposing methods overview
From: Considerations and challenges for sex-aware drug repurposing
Method | Description | Advantages | Disadvantages | Examples |
|---|---|---|---|---|
Data Mining | Analysis of data from various sources (including peer-reviewed published experimental data, databases, screens, pharmaceutical information, EHR’s, etc.) | - Crowdsource data - Multiomic data accessible - Reuse of previously analyzed data | - Limited data for rare diseases and understudied drugs, and dependent on large sample sizes - Inconsistency of data structure - Ethics/privacy (for EHR data) | - Mastermind [89] - Pharos [90] - Iwata H et al. 2015 [91] - Duffy Á et al. 2020 [16] |
Ligand-Binding Prediction | Interactions between ligands and targets are predicted to determine suitable candidates through binding by structural and chemical simulation | - Identify novel drug targets - Identify novel compound structures - Prior knowledge of protein function not required - Detect possible side effects by off-target binding | - Requires target’s tertiary structure - Experimental binding affinities often not recapitulated - Disregards downstream effects - Computationally expensive - Missing biological context to allow tissue or sex-specificity | - Chupakhin V et al. 2013 [92] - Napolitano F et al. 2013 [93] - Vilar S et al. 2014 [94] - Cao R et al. 2014 [95] - Cheng F et al. 2013 [96] |
Molecular Associations | Molecular perturbations are associated with disease, therapeutic outcomes, or drug candidates | - Elucidate drug/disease mechanisms - Compatible with multiomic data - Detect druggable pathways - Exposes off-target drug effects | - High signal-to-noise ratio inhibits deconvolution of signatures - Disregards physiological interactions - Associations may not convey direct causations | - Dr. Insight [97] - signatureSearch [98] - Sanseau P et al. 2012 [99] - Grover MP et al. 2015 [100] |
Networks | The relationship of genes within and between pathways provide insight for upstream and downstream drug targets that may infer treatment for a disease phenotype and/or show drug interactions within a biological system | - Multiomic data - Reveals relationships - Determine mechanistic pathways - Exposes off-target drug effects | - Statistically complex - Computationally expensive - Requires strong signal-to-noise or large datasets to deconvolute signal | - Drug2Ways [101] - Green CS et al. 2015 [102] |
Experimental—Perturbation Screens | Cultured cells are treated with a variety of drugs and screened for phenotypic response | - Shows gene expression as a result of perturbation - Displays consociation between cell receptors and pharmaceuticals - Non-predicted, in-vitro results | - Immortalized cells - Lacks heterogeneity - Limited microenvironment - Costly | - LINCS L1000 profiles [103] - Iljin K et al. 2009 [104] - Shen M et al. 2018 [105] |
Experimental—Binding Assays | The chemical engagement of targets and ligands are tested in vitro to divulge repurposed candidates based on disease-target matching via affinity/thermal stabilization and structures | - Physically measured drug-target binding activity - Captures biophysical features - Reveals promiscuous drug-target interactions | - Disregards downstream effects - Selection of drugs and targets are much more restricted than in silico approaches due to feasibility (cost, time, and accessibility) | - Cellular ThermoStability Assay (CETSA) [106] - Miettinen TP et al. 2014 [107] |
Experimental—Animal Models | Organisms are treated with drugs to model patient response and patient-specific disease-causing genetic variants can be introduced to provide more pertinent system | - Recapitulates full physiological system - Resource for multiomic data collection - In-vivo results - Patient-specific models allow for precision medicine | - Significant financial and time expense - Requires narrowed-down list of candidates - Results frequently do not translate to patient response - Orthologous targets may vary greatly from human target structure | - UAB C-PAM [108] - JAX Center for Precision Genetics [109] - BCM Center for Precision Medicine Models [110] - vivoChip [111] - The Hollow Fiber Model [112] |