Content
Summary
This computational framework integrates drug targets, gene expression, and protein interaction networks to construct signaling networks that reveal a drug's mode of action and prioritize candidates for specific diseases. It also enables personalized safety assessments by analyzing how individual patient cells respond to drugs, specifically through the perturbation of HLA genes associated with adverse reactions.
Intuition
Think of this platform as a sophisticated translator that converts raw biological data into a map of how drugs and diseases interact. By pulling together information from various databases, it builds a 'signaling network' that shows exactly how a drug affects the body's proteins and genes. This map allows researchers to see if a drug's effect aligns with the biological network of a specific disease, helping to rank the best candidates for treatment. Furthermore, by looking at how a drug perturbs specific immune-related genes in a patient's own cells, the system can predict individual safety risks before clinical trials begin.
Derivation
- The platform compiles data from databases like DrugBank, DrugCentral, STITCH, ArrayExpress, and GEO, including drug targets, gene expression profiles, and protein interactions.
- It identifies differentially expressed genes following exposure to a chemical or drug.
- Integer linear programming (ILP) is applied to construct the signaling network, representing the mode of action of a drug candidate or the network of a disease.
- Drug candidates are prioritized by computing the closeness or distance metric between the drug’s signaling network and the disease’s network.
- For personalized medicine, individual patient cells are subjected to a drug for transcriptomic profiling to identify perturbed HLA genes.
- Literature is referenced to assess the risk of severe adverse reactions based on the specific perturbation of these HLA genes.
Worked example
A researcher wants to assess the safety risk of carbamazepine for a specific patient. How does the platform determine this risk?
- Subject the individual patient's cells to carbamazepine for transcriptomic profiling.
- Identify the HLA genes that are significantly perturbed by the drug in these cells.
- Reference literature to assess the risk for severe or fatal adverse drug reactions, such as Stevens-Johnson syndrome, associated with the perturbation of HLA DQB1 genes.
Answer The platform predicts immune-mediated adverse reactions by analyzing the perturbation of specific HLA genes (like HLA DQB1) in the patient's cells and cross-referencing this with known literature on adverse drug reactions.
Common mistakes
- Failing to distinguish between the drug's mode of action (signaling network) and the disease network, which are distinct entities used to calculate closeness.
- Using low-quality data from non-credible databases, which compromises the robustness of the network construction.
- Assuming the platform works without comprehensive prior knowledge of protein interactions and transcription factors.