Multiple simultaneous docking of small molecules in butyrylcholinesterase active site using deep reinforcement learning

master's thesis

• Title (croatian): Višestruko istovremeno dokiranje malih molekula u aktivno mjesto butirilkolinesteraze primjenom dubokog pojačanog učenja
• Author: Filip Sente
• Mentor: Tomica Hrenar (mentor)
• Mentor: Kristian Vlahoviček (komentor)
• Committee member: Tomica Hrenar (predsjednik povjerenstva)
• Committee member: Kristian Vlahoviček (član povjerenstva)
• Committee member: Dubravka Hranilović (član povjerenstva)
• Committee member: Biljana Balen (član povjerenstva)
• Granter: University of Zagreb; Faculty of Science; (Department of Biology); Zagreb
• Defense date and country: 2024-02-27, Croatia
• Scientific / art field, discipline and subdiscipline: NATURAL SCIENCES; Biology

Abstract

Cholinesterase inhibitors are often used in treating Alzheimer’s disease. Molecular docking
represents an important tool in the design of novel inhibitors. In this study, a multicomponent
quantum chemical Monte Carlo semi flexible molecular docking of four selected components of
the Ugi reaction was performed inside the active site of butyrylcholinesterase (BChE). The IC 50
value of compounds was evaluated, and artificial neural networks (ANNs) were trained by deep
reinforcement learning for two of them, using results from docking. Even though components
alone do not inhibit BChE significantly, formaldehyde and benzylamine can be used jointly for
the design of novel drugs because they bind closely to the enzyme’s catalytic triad. In addition,
the advantage of using the three component Ugi reaction is highlighted. Finally, ANN was shown
to be successful in representing a potential energy surface of docking simulations. O btained ANN
can be used as a scoring function for future dockings, with significantly shorter calculation time.

Abstract (croatian)

U liječenju Alzheimerove bolesti često se koriste inhibitori kolinesteraza. Molekularno dokiranje važan je alat u dizajnu novih inhibitora, a u ovome je radu unutar aktivnog mjesta butirilkolinesteraze (BChE) provedeno multikomponentno kvantno-kemijsko Monte Carlo polufleksibilno dokiranje četiriju komponenata Ugijeve reakcije. Određene su im i vrijednosti IC50 te su za dvije od njih istrenirane umjetne neuronske mreže (ANN) primjenom dubokog pojačanog učenja za opis rezultata dokiranja. Iako zasebne molekule ne inhibiraju značajno BChE, formaldehid i benzilamin zajedno mogu poslužiti u dizajnu novih lijekova jer se vežu u blizini katalitičke trijade enzima. Uz to, istaknuta je prednost uporabe trokomponentne Ugijeve reakcije. Naposlijetku, pokazano je da se ANN može uspješno koristiti u reprezentaciji plohe potencijalne energije dokiranja. Dobivena mreža može ubuduće poslužiti kao funkcija bodovanja, uz znatno kraće trajanje proračuna.

• Keywords: Alzheimer’s disease; Ugi reaction; deep learning; artificial neural networks
• Keywords (croatian): Alzheimerova bolest; Ugijeva reakcija; duboko učenje; umjetne neuronske mreže
• Language: english
• URN:NBN: urn:nbn:hr:217:136071
• Project: Number: IP-2022-10-9525; Title: Metom usmjerena sinteza inhibitora kolinesteraza podržana strojnim učenjem; Acronym: TargetCh; Leader: Ines Primožič; Jurisdiction: Croatia; Funder: HRZZ; Funding stream: IP-2022-10
• Study programme: Title: Molecular Biology-University graduate programme; Study programme type: university; Study level: graduate; Academic / professional title: sveučilišni/a magistar/magistra molekularne biologije (sveučilišni/a magistar/magistra molekularne biologije)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Embargoed access; Embargo expiration date: 2026-02-27
• Created on: 2024-03-19 11:34:01