patent policies and offer more incentives from investors to companies focusing on

DR. Additionally, it is very common that safety information of developing drugs,

including repositioned drugs, is lacking at the time of authorization. This is due to

the nature of clinical development, where emphasis is placed on drug efficacy at the

early stages. A more robust risk evaluation plan is highly recommended to address

any safety uncertainties during the post-authorization and post-marketing phase

(Sultana et al. 2020).

In certain cases, some pharmaceutical companies or drug inventors seem inferior

or hesitate to release their compounds/chemical libraries outcomes (e.g. shelved

drugs) to be further explored by other drug developers over its possible applications

or novel indications. This situation stands as one of the barriers to the progress of DR

prospects, especially when the potentially repurposed drug targets diseases that are

beyond the organization’s expertise area. It is also highlighted by Talevi and Bellera

(2020) that regardless of collaborative works between large and smaller firms in the

DR market, it is crucial to establish proper administrative procedures (e.g. agreement

on chemicals/compounds distribution and its subject matter) to ensure benefits to

both parties.

5.4.2

Data Availability and Computational Tools

With the development of high-throughput technology, an enormous amount of

biomedical data has been generated and uploaded on online databases such as

drug-related databases (e.g. DrugBank, PubChem, CTD, and SIDER), disease-

related databases (e.g. Disease Ontology (DO), MalaCards, Online Mendelian

Inheritance in Man (OMIM), and DisGeNET), and protein/gene-related databases

(e.g. UniProtKB, BioGrid, HPRD, and PDB). However, public access to certain

types of valuable and essential information is still very limited. Some of the data may

be missing or insufficient to be processed by modern or classical approaches. The

integration of appropriate databases using computational approaches will enhance

the quality of the analysis and ease the course of identifying new indications for

existing drugs (Talevi and Bellera 2020).

Despite there being various types of computational (based on different

algorithms) and experimental methods, each technique has its own applicability,

drawbacks, and limitations. Thus, none of these methods alone will be able to

decode the complex interaction between drugs, targets, and diseases. Generally,

the most common techniques in computational methods are network- and machine

learning-based for DR (Le and Nguyen-Ngoc 2018; Wang et al. 2020). Neverthe-

less, these techniques only employ a single measurement to analyse the information

similarity of the drugs and disease association in predicting new indications of

existing drugs. In fact, resemblances between the drug-disease interaction are very

multifaceted and must be evaluated from different angles to produce a more precise

quantification of drug-drug and disease-disease relationships. Furthermore, there is

still patchy information about the drug-disease relation in a form of an adjacency

5

Genomic Approaches for Drug Repositioning

65