{"id":{"repo_id":"bu","oai_identifier":"oai:open.bu.edu:2144/53366"},"canonical_url":"https://search.ndltd.org/etd/bu/oai:open.bu.edu:2144/53366","repository":{"repo_id":"bu","name":"Boston University","base_url":"https://open.bu.edu/oai/request"},"display":{"title":"Assessment and combination of methods for causal effect estimation in Mendelian randomization under horizontal pleiotropy","abstract":"Mendelian Randomization (MR) is an application of genetic variation as instrumental variables toward estimating causal relationships between an exposure outcome pair. Instrumental variable frameworks like MR rely on several assumptions including exclusion, where instruments are assumed to affect the outcome only via the exposure. This assumption is frequently violated in real datasets due to genetic pleiotropy. Many extensions to MR have been developed to improve performance in the presence of pleiotropy. This dissertation outlines a simulation framework to generate data suitable for MR and implements commonly used MR methods including inverse variance weighting, MR-Egger, MR-PRESSO, GSMR-Heidi, MR-Median and outlier removal via Studentized residuals or Cook’s distance. These methods are assessed under a variety of pleiotropic and non-pleiotropic scenarios including pleiotropy mediated through confounders. We expand on these common MR techniques by creating ensemble estimators, where features of different methods are combined where appropriate and reassessed. Finally, select methods are further interrogated in the context of our simulated data by varying key parameters which are usually fixed in applied use. Our results illustrate magnitudes of bias, type I error rates, and power of each approach under various pleiotropic settings. Additionally, combining features of methods into ensemble methods does not always result in improved performance. Finally, we provide a framework for practical implementation of MR informed by our results.","abstract_html":"Mendelian Randomization (MR) is an application of genetic variation as instrumental variables toward estimating causal relationships between an exposure outcome pair. Instrumental variable frameworks like MR rely on several assumptions including exclusion, where instruments are assumed to affect the outcome only via the exposure. This assumption is frequently violated in real datasets due to genetic pleiotropy. Many extensions to MR have been developed to improve performance in the presence of pleiotropy. This dissertation outlines a simulation framework to generate data suitable for MR and implements commonly used MR methods including inverse variance weighting, MR-Egger, MR-PRESSO, GSMR-Heidi, MR-Median and outlier removal via Studentized residuals or Cook’s distance. These methods are assessed under a variety of pleiotropic and non-pleiotropic scenarios including pleiotropy mediated through confounders. We expand on these common MR techniques by creating ensemble estimators, where features of different methods are combined where appropriate and reassessed. Finally, select methods are further interrogated in the context of our simulated data by varying key parameters which are usually fixed in applied use. Our results illustrate magnitudes of bias, type I error rates, and power of each approach under various pleiotropic settings. Additionally, combining features of methods into ensemble methods does not always result in improved performance. Finally, we provide a framework for practical implementation of MR informed by our results.","abstract_has_math":false,"creators":["DiCorpo, Daniel A."],"institution":"Boston University","degree_name":"Doctor of Philosophy","degree_level":"doctoral","degree_discipline":"Biostatistics","degree_department":null,"school":null,"contributors":[],"advisors":["Lunetta, Kathryn L."],"committee_chairs":[],"committee_members":[],"year":2025,"date_issued":"2025","date_published":"2025","updated_at":"2026-11-18T20:25:51Z","subjects":["Biostatistics","Genetics","Epidemiology"],"languages":["en_US"],"rights":[],"rights_urls":[],"identifier_entries":[],"person_identifier_entries":[{"key":"dc:identifier.orcid","label":"ORCID","values":["0000-0001-8550-8214"],"render_values":[{"text":"0000-0001-8550-8214","href":"https://orcid.org/0000-0001-8550-8214","code":true}]}]},"links":{"outbound_url":"https://hdl.handle.net/2144/53366","outbound_label":"Handle","outbound_source":"dc:identifier.uri","record_url":"https://hdl.handle.net/2144/53366","record_label":"Repository record","record_kind":"handle","record_source":"dc:identifier.uri","download_url":null,"download_label":null,"download_source":null,"access_status":null,"embargo_until":null,"access_note":null},"source_record":{"url":"https://open.bu.edu/oai/request?verb=GetRecord&metadataPrefix=dim&identifier=oai%3Aopen.bu.edu%3A2144%2F53366","prefix":"dim"},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:contributor.advisor","label":"Advisor","values":["Lunetta, Kathryn L."]},{"key":"dc:creator","label":"Author","values":["DiCorpo, Daniel A."]},{"key":"dc:identifier.orcid","label":"ORCID","values":["0000-0001-8550-8214"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:date.accessioned","label":"Dc Date Accessioned","values":["2026-07-27T17:54:14Z"]},{"key":"dc:date.available","label":"Dc Date Available","values":["2026-07-27T17:54:14Z"]},{"key":"dc:date.issued","label":"Date","values":["2025"]},{"key":"dc:date.updated","label":"Dc Date Updated","values":["2026-07-24T19:15:35Z"]},{"key":"dc:type","label":"Dc Type","values":["Thesis/Dissertation"]},{"key":"thesis:degree_discipline","label":"Discipline","values":["Biostatistics"]},{"key":"thesis:degree_level","label":"Degree Level","values":["doctoral"]},{"key":"thesis:degree_name","label":"Degree Name","values":["Doctor of Philosophy"]},{"key":"thesis:institution_name","label":"Thesis Institution Name","values":["Boston University"]}]},{"id":"subjects_keywords","label":"Subjects and Keywords","entries":[{"key":"dc:subject","label":"Dc Subject","values":["Biostatistics","Genetics","Epidemiology"]}]},{"id":"language_rights","label":"Language and Rights","entries":[{"key":"dc:language.iso","label":"Language (ISO)","values":["en_US"]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier.uri","label":"Identifier URI","values":["https://hdl.handle.net/2144/53366"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description","label":"Description","values":["2025"]},{"key":"dc:description.abstract","label":"Abstract","values":["Mendelian Randomization (MR) is an application of genetic variation as instrumental variables toward estimating causal relationships between an exposure outcome pair. Instrumental variable frameworks like MR rely on several assumptions including exclusion, where instruments are assumed to affect the outcome only via the exposure. This assumption is frequently violated in real datasets due to genetic pleiotropy. Many extensions to MR have been developed to improve performance in the presence of pleiotropy. This dissertation outlines a simulation framework to generate data suitable for MR and implements commonly used MR methods including inverse variance weighting, MR-Egger, MR-PRESSO, GSMR-Heidi, MR-Median and outlier removal via Studentized residuals or Cook’s distance. These methods are assessed under a variety of pleiotropic and non-pleiotropic scenarios including pleiotropy mediated through confounders. We expand on these common MR techniques by creating ensemble estimators, where features of different methods are combined where appropriate and reassessed. Finally, select methods are further interrogated in the context of our simulated data by varying key parameters which are usually fixed in applied use. Our results illustrate magnitudes of bias, type I error rates, and power of each approach under various pleiotropic settings. Additionally, combining features of methods into ensemble methods does not always result in improved performance. Finally, we provide a framework for practical implementation of MR informed by our results."]},{"key":"dc:title","label":"Title","values":["Assessment and combination of methods for causal effect estimation in Mendelian randomization under horizontal pleiotropy"]}]}],"canonical_facts":{"dc:contributor.advisor":["Lunetta, Kathryn L."],"dc:creator":["DiCorpo, Daniel A."],"dc:date.accessioned":["2026-07-27T17:54:14Z"],"dc:date.available":["2026-07-27T17:54:14Z"],"dc:date.issued":["2025"],"dc:date.updated":["2026-07-24T19:15:35Z"],"dc:description":["2025"],"dc:description.abstract":["Mendelian Randomization (MR) is an application of genetic variation as instrumental variables toward estimating causal relationships between an exposure outcome pair. Instrumental variable frameworks like MR rely on several assumptions including exclusion, where instruments are assumed to affect the outcome only via the exposure. This assumption is frequently violated in real datasets due to genetic pleiotropy. Many extensions to MR have been developed to improve performance in the presence of pleiotropy. This dissertation outlines a simulation framework to generate data suitable for MR and implements commonly used MR methods including inverse variance weighting, MR-Egger, MR-PRESSO, GSMR-Heidi, MR-Median and outlier removal via Studentized residuals or Cook’s distance. These methods are assessed under a variety of pleiotropic and non-pleiotropic scenarios including pleiotropy mediated through confounders. We expand on these common MR techniques by creating ensemble estimators, where features of different methods are combined where appropriate and reassessed. Finally, select methods are further interrogated in the context of our simulated data by varying key parameters which are usually fixed in applied use. Our results illustrate magnitudes of bias, type I error rates, and power of each approach under various pleiotropic settings. Additionally, combining features of methods into ensemble methods does not always result in improved performance. Finally, we provide a framework for practical implementation of MR informed by our results."],"dc:identifier.orcid":["0000-0001-8550-8214"],"dc:identifier.uri":["https://hdl.handle.net/2144/53366"],"dc:language.iso":["en_US"],"dc:subject":["Biostatistics","Genetics","Epidemiology"],"dc:title":["Assessment and combination of methods for causal effect estimation in Mendelian randomization under horizontal pleiotropy"],"dc:type":["Thesis/Dissertation"],"thesis:degree_discipline":["Biostatistics"],"thesis:degree_level":["doctoral"],"thesis:degree_name":["Doctor of Philosophy"],"thesis:institution_name":["Boston University"]},"updated_at":"2026-11-18T20:25:51Z"}