Kvantno-kemijski proračuni i strojno učenje u predviđanju bioaktivnosti novih spojeva

doctoral thesis

• Title (english): Quantum chemical calculations and machnie learning in predicting bioactivity of new compounds
• Author: Ana Mikelić
• Mentor: Tomica Hrenar (mentor)
• Committee member: Gordan Horvat (predsjednik povjerenstva)
• Committee member: Borislav Kovačević (član povjerenstva)
• Committee member: Aleksandra Maršavelski (član povjerenstva)
• Granter: University of Zagreb; Faculty of Science; (Department of Chemistry); Zagreb
• Defense date and country: 2024-05-24, Croatia
• Scientific / art field, discipline and subdiscipline: NATURAL SCIENCES; Chemistry
• Universal decimal classification (UDC): 54 - Chemistry. Crystallography. Mineralogy

Abstract

Male molekule mogu interagirati s biološkim makromolekulama i tako utjecati na homeostazu organizma. U okviru ovoga doktorskog rada računalno su istraženi odabrani kemijski spojevi kao potencijalni antimikrobni agensi i/ili inhibitori kolinesteraza. Najvažnije interakcije malih molekula s makromolekulama određene su metodama molekularnog dokiranja: 1. novorazvijena metoda tenzorske dekompozicije trajektorija ab initio molekularne dinamike te 2. kvantno-kemijsko Monte Carlo dokiranje unutar aktivnog mjesta enzima. Molekularno dokiranje ab initio molekularnom dinamikom uključuje protokol strojnog učenja za određivanje duljine trajanja simulacija kada su uzorkovani potpuni konfiguracijski prostori. Kvantno-kemijsko Monte Carlo dokiranje podrazumijeva uzorkovanje struktura putem generiranja slučajnih konfiguracija sustava (aktivno mjesto enzima i mala molekula) te kvantno-kemijske proračune energije. Ekstenzivnom primjenom strojnog učenja uspostavljeni su najbolji mogući multivarijatni linearni regresijski modeli koji povezuju biološku aktivnost promatranih spojeva s njihovim teorijski uzorkovanim plohama potencijalne energije. Dobiveni rezultati predstavljat će osnovu za daljnja istraživanja najvažnijih interakcija malih molekula s kolinesterazama i za predviđanje bioaktivnosti te "pametan" dizajn novih bioaktivnih molekula.

Abstract (english)

Small molecules can interact with biological macromolecules and thus affect the organism's homeostasis. Within this doctoral dissertation, selected chemical compounds were theoretically investigated as potential antimicrobial agents and/or cholinesterase inhibitors. The most important interactions of small molecules with macromolecules were determined by molecular docking methods: 1. newly developed method of tensor decomposition of ab initio molecular dynamics trajectories and 2. quantum-chemical Monte Carlo docking within the active site of the enzyme. Molecular docking by ab initio molecular dynamics includes a machine learning protocol for determining the length of simulations where complete configuration spaces are sampled. Quantum-chemical Monte Carlo docking implies sampling of structures through the generation of random system configurations (enzyme active site and small molecule) and quantum-chemical energy calculations. Through the extensive application of machine learning, the best possible multivariate linear regression models that connect the biological activity of the observed compounds with their theoretically sampled potential energy surfaces were established. The obtained results will represent the basis for further research of the most important interactions of small molecules with cholinesterases as well as the prediction of bioactivity and the "smart" design of new bioactive molecules

• Keywords: kvantno-kemijski proračuni; molekularno dokiranje; strojno učenje; regresijski modeli; derivati kinuklidina; inhibitori kolinesteraza; antimikrobna aktivnost
• Keywords (english): quantum chemical calculations; molecular docking; machine learning; regression models; quinuclidine derivatives; cholinesterase inhibitors; antimicrobial activity
• Language: croatian
• URN:NBN: urn:nbn:hr:217:712062
• Promotion: 2024
• Project: Number: IP-2016-06-3775; Title: Aktivnošću i in silico usmjeren dizajn malih bioaktivnih molekula; Title: Activity and in silico guided design of bioactive small molecules; Acronym: ADESIRE; Leader: Tomica Hrenar; Jurisdiction: Croatia; Funder: Hrvatska zaklada za znanost; Funding stream: Research Projects
• Study programme: Title: Doctoral study; Study programme type: university; Study level: postgraduate; Academic / professional title: doktor/doktorica u području prirodnih znanosti (doktor/doktorica u području prirodnih znanosti)
• Type of resource: Text
• Extent: 144, XIX str. ; 30 cm
• File origin: Born digital
• Access conditions: Embargoed access; Embargo expiration date: 2026-05-24
• Created on: 2024-05-29 12:22:02