PREDICT: a method for inferring novel drug indications with application to personalized medicine

Abstract

Inferring potential drug indications, for either novel or approved drugs, is a key step in drug development. Previous computational methods in this domain have focused on either drug repositioning or matching drug and disease gene expression profiles. Here, we present a novel method for the large-scale prediction of drug indications (PREDICT) that can handle both approved drugs and novel molecules. Our method is based on the observation that similar drugs are indicated for similar diseases, and utilizes multiple drug-drug and disease-disease similarity measures for the prediction task. On cross-validation, it obtains high specificity and sensitivity (AUC=0.9) in predicting drug indications, surpassing existing methods. We validate our predictions by their overlap with drug indications that are currently under clinical trials, and by their agreement with tissue-specific expression information on the drug targets. We further show that disease-specific genetic signatures can be used to accurately predict drug indications for new diseases (AUC=0.92). This lays the computational foundation for future personalized drug treatments, where gene expression signatures from individual patients would replace the disease-specific signatures.

Figure 1

Algorithmic pipeline: formation of drug–disease associations (A), creation of drug–drug and disease–disease similarity metrics (B), scoring possible drug indications according to their similarity to known drug indications (C) and integration of the similarities to classification features and subsequent classification (D).

Figure 2

Validation scheme for drug repositioning predictions. We identify a score cutoff that yields the best P-value against drug indication originating from single textual indication source (low confidence) (A). Applying the cutoff, we validate the selected top ranking predictions against indications under test in clinical trials (B) and the co-occurrence of drug targets and indicated diseases in the same tissues (C).

Figure 3

Distributions of Anatomic, Therapeutic and Chemical (ATC) top level classes among drug–disease associations in the gold standard (A) and the predicted associations (B). The relative ratio between the two distributions for each ATC class is shown in subfigure (C). ATC classes include: alimentary tract and metabolism (A), blood and blood forming organs (B), cardiovascular system (C), dermatologicals (D), genito urinary system and sex hormones (G), antineoplastic and immunomodulating agents (L), musculo-skeletal system (M), nervous system (N), respiratory system (R) and sensory organs (S).