Link to the University of Pittsburgh Homepage
Link to the University Library System Homepage Link to the Contact Us Form

Applications of ML/DL overcoming current challenges in structure-based drug design

Ji, Beihong (2024) Applications of ML/DL overcoming current challenges in structure-based drug design. Doctoral Dissertation, University of Pittsburgh. (Unpublished)

Primary Text

Download (13MB) | Preview


In the rapidly evolving landscape of drug discovery and development, the application of structure-based drug design (SBDD) methods has become increasingly crucial. Despite the advancements in this field, several challenges continue to impede progress, particularly in the areas of binding heterogeneity, targeted ligand development, and the exploration of the vast chemical space for drug-like molecules. This dissertation addresses these challenges by integrating traditional SBDD techniques with advanced machine learning (ML) or deep learning (DL) methodologies, paving the way for more efficient and targeted drug discovery processes.
The research began with a focus on cannabinoid receptors, employing a set of molecular modeling methods, to predict binding affinity and selectivity of compounds. This approach aids in identifying key interaction sites, facilitating the development of potent and selective ligands. The study also explored complexities of drug-drug interactions, providing insights into the dynamic interplay of pharmacodynamic and pharmacokinetic factors.
To combat the issue of binding heterogeneity, novel computational approaches were introduced. One involves the creation of an advanced computational algorithm that combines structure-based docking scores with ligand-based structural similarities. Another is the development of an ML-based scoring function using ligand-residue interaction profiles. These innovative methods have shown a marked improvement in the accuracy of binding affinity predictions compared to traditional techniques.
A function-based screening methodology was also presented to predict the functionality of CB2 ligands. The success of this approach in the development of CB2 agonists, with a significant success rate, demonstrates its potential in overcoming challenges in ligand design and fulfilling specific biological functions.
Additionally, the dissertation introduced DRUG-GAN, a deep learning model for the de novo generation of molecular structures. This model, when integrated with similarity search methods, significantly narrows the chemical search space and accelerates the process of lead compound identification in the drug discovery and development.
In conclusion, this research successfully demonstrates the effectiveness of combining computational modeling techniques with artificial intelligence in addressing key challenges in SBDD. The methodologies developed herein not only enhance the precision of drug design but also expand the horizons for discovering novel drug candidates, setting a new benchmark in the field of pharmaceutical research.


Social Networking:
Share |


Item Type: University of Pittsburgh ETD
Status: Unpublished
CreatorsEmailPitt UsernameORCID
Ji, Beihongbej22@pitt.edubej220000-0003-0387-4056
ETD Committee:
TitleMemberEmail AddressPitt UsernameORCID
Committee ChairWang, JunmeiJUW79@pitt.eduJUW79
Committee CoChairGibbs,
Committee MemberPoloyac,
Committee MemberKoes,
Committee MemberWang, LirongLIW30@pitt.eduLIW30
Thesis AdvisorWang, JunmeiJUW79@pitt.eduJUW79
Date: 29 February 2024
Date Type: Publication
Defense Date: 18 December 2023
Approval Date: 29 February 2024
Submission Date: 31 January 2024
Access Restriction: No restriction; Release the ETD for access worldwide immediately.
Number of Pages: 238
Institution: University of Pittsburgh
Schools and Programs: School of Pharmacy > Pharmaceutical Sciences
Degree: PhD - Doctor of Philosophy
Thesis Type: Doctoral Dissertation
Refereed: Yes
Uncontrolled Keywords: computer-aided drug design, CADD, structure-based drug design, SBDD, machine learning, ML
Date Deposited: 29 Feb 2024 16:55
Last Modified: 29 Feb 2024 16:55


Monthly Views for the past 3 years

Plum Analytics

Actions (login required)

View Item View Item