Obtain MEDLINE abstracts from the protein list’s construction.

Construction of protein list Retrieve MEDLINE abstracts and identify drug terms relevant to protein list Plasmodium falciparum was used as a query to download the title and abstracts from the MEDLINE database through PubMed. The abstracts for the previous five years (8036) were downloaded locally. Text data mining process including tokenization, stop word removal stemming were carried out. 1554 abstracts representing the relevant abstracts for each protein were retrieved. Chemical and drug category terms were downloaded from the MeSH database (Figure 3), which is divided into 16 categories [27], in order to locate drug terms that are relevant to the protein list. All terms of MeSH database were searched with the 1554 abstracts and 80 drug/MeSH terms out of 292 were identified. The occurrence of the identified drug terms for P. falciparum parasite is shown in Table 2. In 1554 abstracts, the inorganic chemicals were mentioned 373 times, indicating that the majority of the P. falciparum drugs reported are inorganic chemicals. A two-dimensional matrix in which protein names and drug terms are listed in rows and columns, respectively, can be used to construct the association between the protein list and the identified drug terms. The matrix is filled with the score obtained using the equation (1). Based on the positive and negative values of the score dtp, the protein-drug pair is over- or under-represented, respectively. So, the higher score is more significant in the matrix