Biomedical Text Link Prediction for Drug Discovery: A Case Study with COVID-19

Abstract

Link prediction in artificial intelligence is used to identify missing links or derive future relationships that can occur in complex networks. A link prediction model was developed using the complex heterogeneous biomedical knowledge graph, SemNet, to predict missing links in biomedical literature for drug discovery. A web application visualized knowledge graph embeddings and link prediction results using TransE, CompleX, and RotatE based methods. The link prediction model achieved up to 0.44 hits@10 on the entity prediction tasks. The recent outbreak of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), also known as COVID-19, served as a case study to demonstrate the efficacy of link prediction modeling for drug discovery. The link prediction algorithm guided identification and ranking of repurposed drug candidates for SARS-CoV-2 primarily by text mining biomedical literature from previous coronaviruses, including SARS and middle east respiratory syndrome (MERS). Repurposed drugs included potential primary SARS-CoV-2 treatment, adjunctive therapies, or therapeutics to treat side effects. The link prediction accuracy for nodes ranked highly for SARS coronavirus was 0.875 as calculated by human in the loop validation on existing COVID-19 specific data sets. Drug classes predicted as highly ranked include anti-inflammatory, nucleoside analogs, protease inhibitors, antimalarials, envelope proteins, and glycoproteins. Examples of highly ranked predicted links to SARS-CoV-2: human leukocyte interferon, recombinant interferon-gamma, cyclosporine, antiviral therapy, zidovudine, chloroquine, vaccination, methotrexate, artemisinin, alkaloids, glycyrrhizic acid, quinine, flavonoids, amprenavir, suramin, complement system proteins, fluoroquinolones, bone marrow transplantation, albuterol, ciprofloxacin, quinolone antibacterial agents, and hydroxymethylglutaryl-CoA reductase inhibitors. Approximately 40% of identified drugs were not previously connected to SARS, such as edetic acid or biotin. In summary, link prediction can effectively suggest repurposed drugs for emergent diseases.

A Systematic Review on the Contribution of Artificial Intelligence in the Development of Medicines for COVID-2019

“…Among the most ranked drug classes were anti-inflammatory, nucleoside analogues, protease inhibitors, antimalarials, envelope proteins, and glycoproteins. For instance, highly ranked predicted links to SARS-CoV-2 were identified for human leukocyte interferon, recombinant interferon-gamma ( a Type 1 inflammatory cytokine; antitumor properties, antiviral activities, and important immunoregulatory functions ; DrugBank), cyclosporine ( a steroid-sparing immunosuppressant used in organ and bone marrow transplants as well as inflammatory conditions such as ulcerative colitis, rheumatoid arthritis, and atopic dermatitis ; DrugBank), antiviral therapy, zidovudine ( a dideoxynucleoside used in the treatment of HIV infection ; DrugBank), chloroquine (antimalarial drug, see Appendix A ), methotrexate ( an antineoplastic agent used the treatment of a wide variety of cancers as well as severe psoriasis, severe rheumatoid arthritis, and juvenile rheumatoid arthritis ; DrugBank), artemisinin ( treatment of malaria ; DrugBank), alkaloids, glycyrrhizic acid (it is extracted from the root of the licorice plant; Glycyrrhiza glabra; hepatoprotective drug in cases of chronic hepatitis ; DrugBank), quinine ( an alkaloid derived from the bark of the cinchona tree; it is used as an antimalarial drug ; DrugBank), flavonoids, amprenavir ( a protease inhibitor used to treat HIV infection ; DrugBank), suramin ( a polyanionic compound used in the treatment of African trypanosomiasis, with potent antineoplastic properties; DrugBank), complement system proteins, fluoroquinolones, albuterol ( a beta-2 adrenergic receptor agonist used to treat asthma, bronchitis, COPD, as well as prevent exercise induced bronchospasms ; DrugBank), ciprofloxacin ( a second generation fluoroquinolone used to treat various susceptible bacterial infections ; DrugBank), quinolone antibacterial agents, and hydroxymethylglutaryl-CoA reductase inhibitors [ 39 ].…”

“…Overall, diverse new repurposed drugs against the treatment of COVID-2019 were suggested in this systematic review through AI techniques [ 34 , 35 , 36 , 37 , 38 , 39 , 40 , 41 ], such as: valrubicin, aprepitant, dihydroergotamine, * bivalirudin, eltrombopag, eribulin, fulvestrant, idarubicin, ivermectin, ledipasvir, lifitegrast, nystatin, regorafenib, trypan blue, vitamin E, ruxolitinib, nafcillin, nabumetone, octacosanol, cinametic acid, trabectedin, lauric acid, acorbyl palmitrate, palmidrol, salmeterol, simvastatin, guaifenesin, verteporfin, metergoline, rescinnamine, leuprolide, lusutrombopag, telotristat, fostamatinib, tofacitinib, etoricoxib, ziprasidone, interferon-gamma; cyclosporine, zidovudine, methotrexate, artemisinin, glycyrrhizin acid, quinine, suramin, albuterol, ciprofloxacin, spirapril, lisinopril, and captopril. …”

“…Overall, diverse new repurposed drugs against the treatment of COVID-2019 were suggested in this systematic review through AI techniques [34][35][36][37][38][39][40][41], such as:…”

CompositeView: A Network-Based Visualization Tool

“…Here we develop CompositeView, an open-source Python-based data visualization tool, to assist domain specialists in deriving actionable insights from large, complex data sets that can be visualized in network form. CompositeView was originally inspired by link prediction and relevance scoring methods [ 6 ], but was generalized for a greater breadth of data. The key gap filled by CompositeView is the ability to automatically calculate and update composite scores as the user simultaneously interacts with the data.…”

When Cyclodextrins Met Data Science: Unveiling Their Pharmaceutical Applications through Network Science and Text-Mining

“…The increasing availability of data in different scientific fields conceives significant opportunities to leverage the data to guide research [ 26 , 27 , 28 , 29 , 30 , 31 , 32 , 33 ]. Thus, data-driven approaches based on methods similar to those presented herein [ 18 , 34 , 35 ], along with recent discoveries in machine learning, natural language processing, and text-mining, can pave the way for scientific discoveries and innovation in the pharmaceutical field for both academia and industry.…”