{"id":{"repo_id":"msu","oai_identifier":"oai:d.lib.msu.edu:etd_53433"},"canonical_url":"https://search.ndltd.org/etd/msu/oai:d.lib.msu.edu:etd_53433","repository":{"repo_id":"msu","name":"Michigan State University","base_url":"https://d.lib.msu.edu/oai"},"display":{"title":"STRUCTURALLY INFORMED DEEP LEARNING MODELS FOR PROTEIN-LIGAND INTERACTION PREDICTION","abstract":"Thesis (Ph.D.)--Michigan State University. Chemical Engineering - Doctor of Philosophy, 2026","abstract_html":"Thesis (Ph.D.)--Michigan State University. Chemical Engineering - Doctor of Philosophy, 2026","abstract_has_math":false,"creators":["Eaves, Joelle N"],"institution":null,"degree_name":null,"degree_level":null,"degree_discipline":null,"degree_department":null,"school":null,"contributors":["Woldring, Daniel R","Hickey, David","Orlando, Benjamin","Calabrese Barton, Scott"],"advisors":[],"committee_chairs":[],"committee_members":[],"year":2026,"date_issued":"2026","date_published":"2026","updated_at":"2026-11-18T18:59:53Z","subjects":["Computational chemistry","Biochemistry","Biophysics","Chemistry--Computer simulation"],"languages":["eng"],"rights":["Attribution 4.0 International"],"rights_urls":[],"identifier_entries":[{"key":"dc:identifier","label":"Identifier","values":["etd:53433","local:Eaves_grad.msu_0128D_22502"],"render_values":[{"text":"etd:53433","href":null,"code":true},{"text":"local:Eaves_grad.msu_0128D_22502","href":null,"code":true}]}],"person_identifier_entries":[]},"links":{"outbound_url":"https://doi.org/doi:10.25335/e596-ag53","outbound_label":"DOI","outbound_source":"dc:identifier","record_url":"https://doi.org/doi:10.25335/e596-ag53","record_label":"Repository record","record_kind":"doi","record_source":"dc:identifier","download_url":null,"download_label":null,"download_source":null,"access_status":null,"embargo_until":null,"access_note":null},"source_record":{"url":"https://d.lib.msu.edu/oai?verb=GetRecord&metadataPrefix=oai_dc&identifier=oai%3Ad.lib.msu.edu%3Aetd_53433","prefix":"oai_dc"},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:contributor","label":"Contributor","values":["Woldring, Daniel R","Hickey, David","Orlando, Benjamin","Calabrese Barton, Scott"]},{"key":"dc:creator","label":"Author","values":["Eaves, Joelle N"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:date","label":"Dc Date","values":["2026"]},{"key":"dc:relation","label":"Dc Relation","values":["Electronic Theses & Dissertations"]},{"key":"dc:type","label":"Dc Type","values":["text","Electronic dissertations"]}]},{"id":"subjects_keywords","label":"Subjects and Keywords","entries":[{"key":"dc:subject","label":"Dc Subject","values":["Computational chemistry","Biochemistry","Biophysics","Chemistry--Computer simulation"]}]},{"id":"language_rights","label":"Language and Rights","entries":[{"key":"dc:language","label":"Dc Language","values":["eng"]},{"key":"dc:rights","label":"Dc Rights","values":["Attribution 4.0 International"]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier","label":"Identifier","values":["etd:53433","local:Eaves_grad.msu_0128D_22502","https://doi.org/doi:10.25335/e596-ag53"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description","label":"Description","values":["Thesis (Ph.D.)--Michigan State University. Chemical Engineering - Doctor of Philosophy, 2026","Structure-based drug discovery has been transformed by rapid advances in protein structuredetermination and prediction. Cryo-EM now captures membrane proteins at near-atomic resolution. Methods such as AlphaFold3 enable accurate protein-ligand co-folding at scale. Together, these developments have dramatically expanded the structural data available for computational protein ligand binding affinity prediction (PLBAP) and opened new opportunities for previously difficult to-model targets. This dissertation develops and critically evaluates structure-based deep learning models for small molecule interaction prediction with three interconnected contributions. Chapter 2 introduces HOLIgraph, the first structure-based deep learning model for predicting small molecule inhibitors of OATP1B1, a key drug transporter protein. Enabled by recent cryo-EM structures of OATP1B1 in multiple conformational states, HOLIgraph encodes protein-ligand interaction data from ensemble docking simulations into a heterogeneous graph neural network. HOLIgraph achieves a median balanced accuracy exceeding 90%, significantly outperforming ligand-only models. Chapter 3 examines how amenable computationally generated structures are to practical PLBAP workflows. Because experimental co-crystal structures are rarely available in prospective drug discovery, PLBAP models are routinely applied to docked or predicted inputs. However, the performance consequences have not been broadly quantified prior to our study. We evaluated five reproducible PLBAP pipelines across crystal structures, rigid-receptor docking, and AlphaFold3 co-folding to rigorously quantify the benchmark-to-deployment mismatch. Performance consistently degraded under computed inputs. Chapter 4 audits the reproducibility of 50 modern PLBAP models. We document that only 34% of models provide sufficient materials to be rerun. Worse, performance was only reproducible for 16%ofmodels. To address this, we propose a five-item checklist of reproducibility-critical materials to be used as a lightweight community standard for future PLBAP releases.","Description based on online resource. Title from PDF t.p. (Michigan State University Fedora Repository, viewed ).","Includes bibliographical references."]},{"key":"dc:format","label":"Dc Format","values":["application/pdf","1 online resource (174 p.) : ill."]},{"key":"dc:title","label":"Title","values":["STRUCTURALLY INFORMED DEEP LEARNING MODELS FOR PROTEIN-LIGAND INTERACTION PREDICTION"]}]}],"canonical_facts":{"dc:contributor":["Woldring, Daniel R","Hickey, David","Orlando, Benjamin","Calabrese Barton, Scott"],"dc:creator":["Eaves, Joelle N"],"dc:date":["2026"],"dc:description":["Thesis (Ph.D.)--Michigan State University. Chemical Engineering - Doctor of Philosophy, 2026","Structure-based drug discovery has been transformed by rapid advances in protein structuredetermination and prediction. Cryo-EM now captures membrane proteins at near-atomic resolution. Methods such as AlphaFold3 enable accurate protein-ligand co-folding at scale. Together, these developments have dramatically expanded the structural data available for computational protein ligand binding affinity prediction (PLBAP) and opened new opportunities for previously difficult to-model targets. This dissertation develops and critically evaluates structure-based deep learning models for small molecule interaction prediction with three interconnected contributions. Chapter 2 introduces HOLIgraph, the first structure-based deep learning model for predicting small molecule inhibitors of OATP1B1, a key drug transporter protein. Enabled by recent cryo-EM structures of OATP1B1 in multiple conformational states, HOLIgraph encodes protein-ligand interaction data from ensemble docking simulations into a heterogeneous graph neural network. HOLIgraph achieves a median balanced accuracy exceeding 90%, significantly outperforming ligand-only models. Chapter 3 examines how amenable computationally generated structures are to practical PLBAP workflows. Because experimental co-crystal structures are rarely available in prospective drug discovery, PLBAP models are routinely applied to docked or predicted inputs. However, the performance consequences have not been broadly quantified prior to our study. We evaluated five reproducible PLBAP pipelines across crystal structures, rigid-receptor docking, and AlphaFold3 co-folding to rigorously quantify the benchmark-to-deployment mismatch. Performance consistently degraded under computed inputs. Chapter 4 audits the reproducibility of 50 modern PLBAP models. We document that only 34% of models provide sufficient materials to be rerun. Worse, performance was only reproducible for 16%ofmodels. To address this, we propose a five-item checklist of reproducibility-critical materials to be used as a lightweight community standard for future PLBAP releases.","Description based on online resource. Title from PDF t.p. (Michigan State University Fedora Repository, viewed ).","Includes bibliographical references."],"dc:format":["application/pdf","1 online resource (174 p.) : ill."],"dc:identifier":["etd:53433","local:Eaves_grad.msu_0128D_22502","https://doi.org/doi:10.25335/e596-ag53"],"dc:language":["eng"],"dc:relation":["Electronic Theses & Dissertations"],"dc:rights":["Attribution 4.0 International"],"dc:subject":["Computational chemistry","Biochemistry","Biophysics","Chemistry--Computer simulation"],"dc:title":["STRUCTURALLY INFORMED DEEP LEARNING MODELS FOR PROTEIN-LIGAND INTERACTION PREDICTION"],"dc:type":["text","Electronic dissertations"]},"updated_at":"2026-11-18T18:59:53Z"}