**6.1** *De novo* **structure-based prediction**

Based only on drug structure, this approach is useful for virtual screening of large compound libraries. This approach has advantage to provide structural

**Figure 4.** *Approaches (experimental and computational) used in drug repurposing.*

*Antituberculosis Drug Repurposing: A New Hope for Tackling Multi-Challenging TB in Timely… DOI: http://dx.doi.org/10.5772/intechopen.101642*

insights about the interaction and further guide the optimization the structure to improve the binding affinity for its target. The approach is computationally much demanding, limiting its large-scale use, many-to-many DTI prediction tasks. In a *ligand-based* approach, constructing a "pseudo-drug" representation called a *pharmacophore* model is used for elucidating the interaction with the chosen target [65]. Pharmacophore models can be constructed from analysis of the target's binding pocket, or derived using a set of positive and negative examples of compounds interacting with the target. Compared with molecular docking, this approach is more computationally efficient and has better accuracy [66]. **Figure 4** summarizes various approaches for drug repurposing.

Different computational approaches can be used independently or in amalgamation to systematically analyze different types of large-scale data to obtain significant interpretations for repurposing hypothesis. Experimental approaches can also be used to recognize repurposing opportunities [67]. Computational approaches are mainly datadriven; they involve systematic analysis of data of any type (such as gene expression, chemical structure, genotype or proteomic data, or electronic health records (EHRs)), which can then lead to the formulation of repurposing hypo-theses.
