Inference of Gene Regulatory Networks for Drug Repurposing: A CLR-Based Strategy to Predict Transcriptional Drug Effects

<p xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" class="first" dir="auto" id="d9292592e148">Drug repurposing provides a cost-effective approach to identifying new therapeutic indications for existing compounds. Here, we present a computational pipeline based on gene expression data and network inference to predict the transcriptional effects of drugs. Using the Context Likelihood of Relatedness (CLR) algorithm, we construct gene regulatory networks from normalized transcriptomic profiles. The resulting network captures non-linear dependencies among genes by computing mutual information and contextual z-scores, enhancing the specificity of inferred gene-gene relationships. <p xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" dir="auto" id="d9292592e150">We leverage this network to model the propagation of drug-induced perturbations across gene modules, enabling the identification of likely downstream transcriptional effects. This approach facilitates the prioritization of candidate drugs whose impact aligns with or counteracts disease-specific gene expression signatures. Applications include the prediction of synergistic drug pairs, the identification of unexpected targets, and the generation of mechanistic hypotheses. <p xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" dir="auto" id="d9292592e152">Preliminary validations using reference compound libraries show that the CLR-inferred network outperforms correlation-based methods in recapitulating known drug-gene interactions. This method is being integrated into a scalable framework for systematic drug repurposing across multiple disease contexts.