Uporaba računalniške analize znanstvene literature za iskanje novih snovi s potencialom za anabolično delovanje : enovit magistrski študij farmacija

Žvegelj, Gal (author), Žiberna, Lovro (mentor), Hristovski, Dimitar (comentor)

Doping pomeni v športu resno težavo, saj lahko športniki z zlorabo prepovedanih snovi in/ali postopkov pridobijo nepošteno prednost pred sotekmovalci in resno ogrozijo svoje zdravje. Naš cilj je bil razviti in ovrednotiti računalniško metodo za odkrivanje novih snovi, ki neposredno ali posredno delujejo anabolično in imajo potencial, da bi jih zlorabili za doping, saj so v športu te snovi najpogosteje zlorabljene. Za vzpostavitev našega eksperimentalnega modela smo uporabili vse snovi z anaboličnim delovanjem, ki so trenutno v športu prepovedane in uvrščene na Listo prepovedanih snovi in postopkov (LPSP), ki jo vsako leto izda Svetovna protidopinška agencija. Naša metoda spada med nove računalniške pristope k analizi biomedicinske znanstvene literature (angl. Literature-based discovery), ki omogočajo pridobivanje novih hipotez in/ali teorij z obdelavo že obstoječega znanja. Pri raziskovalnem delu smo uporabili program SemBT, ki omogoča analizo semantičnih relacij. Za pridobitev želenih rezultatov smo morali sestaviti iskalne nize, pri čemer smo morali dobro poznati farmakološke mehanizme delovanja iskanih snovi. Semantične relacije in rezultate smo grafično obdelali v programskem okolju Neo4J. Z uporabo istih iskalnih nizov smo v programskem orodju VOS Viewer naredili tudi bibliometrično analizo ustvarjenih omrežij, kjer smo primerjali objave znanstvenih člankov s citiranostjo. Uporabljeni program SemBT je še vedno v prototipni fazi, zato smo dobljene rezultate tudi ovrednotili na več načinov. Ocenili smo zmožnost in učinkovitost programa za: i. odkrivanje že znanih prepovedanih snovi z Liste prepovedanih snovi in postopkov (program je našel 58 % snovi z LPSP); ii. zmožnost programa za odkrivanje novih snovi z možnostjo zlorabe za namene dopinga (program je našel 207 novih snovi, ki jih ni na Listi) in iii. pravilnost dobljenih rezultatov (70,2 % rezultatov je bilo pravilnih). Ocenjujemo, da ima semantični pristop odkrivanja novih snovi s potencialom za zlorabo v športu velike možnosti za učinkovitost v boju proti dopingu.

• Language: Slovenian
• Keywords: anabolično delovanje anabolični androgeni steroidi telesna zmogljivost orodje SemBT
• Work type: Master's thesis/paper
• Typology: 2.09 - Master's Thesis
• Organization: FFA - Faculty of Pharmacy
• Place of publishing: Ljubljana
• Publisher: [G. C. Žvegelj]
• Year: 2018
• Number of pages: VI, 65 str.
• PID: 20.500.12556/RUL-120366
• UDC: 796.011.5:178.8+547.92(043.3)
• COBISS.SI-ID: 4657265
• Publication date in RUL: 18.09.2020

Secondary language

• Language: English
• Title: Literature based discovery for the indetification of novel substances with the potential for anabolic activity

Abstract

Doping in sport presents a serious problem. Athletes can gain unfair competitive advantage over the others, as well as doping use can impose serious health risks. Our goal was to develop and evaluate a computational method for the identification of novel substances with direct or indirect anabolic effects and consequential potential for doping abuse. Anabolic agents are the most frequently abused substances in sport. For the construction of our experimental model we picked out all substances with anabolic activity from the List of prohibited substances and methods, which is edited and updated annualy by World Anti-Doping Agency. Our method is a novel computational approach for analysing biomedical literature (literature-based discovery) with the aim to generate novel hypotheses from the existing bulk academic knowledge. We used software SemBT to define and analyze semantic relations. We constructed several search queries based on comprehensive knowledge about pharmacological activity of prohibited substances. Obtained results and semantic relations were graphically analysed with Neo4J software. Same search queries were also used in the software tool VOS Viewer to perform bibliometric analysis of the obtained networks based on bibliographic coupling and citations. Since SemBT software is stil in prototype phase we evaluated the obtained results in several manners. We assesed the software's efficacy to: i. identify substances already on the List of prohibited substances and methods (software identified 58 % of substances mentioned on the List); ii. to identify novel substances with the doping potential abuse (software identified 207 novel substances, not currently included on the List) and iii. the level of results correctness (70,2 % of results were correct).

• Keywords: doping anabolism anabolic androgenic steroids increased performance semantic tool SemBT