{"id":{"repo_id":"cambridge","oai_identifier":"oai:www.repository.cam.ac.uk:1810/383355"},"canonical_url":"https://search.dev.ndltd.org/etd/cambridge/oai:www.repository.cam.ac.uk:1810/383355","repository":{"repo_id":"cambridge","name":"Cambridge University","base_url":"https://api.repository.cam.ac.uk/server/oai/request"},"display":{"title":"Accurate estimation of drug combination synergies with uncertainty quantification","abstract":"In pharmacology, there exists a range of different quantification frameworks to estimate the synergistic effect of drug combinations. However, the varying assumptions and the absence of a universal gold standard across these frameworks make it difficult to quantitatively prioritise which drug combinations from large screening experiments should advance for further investigation. Furthermore, existing frameworks lack accurate uncertainty quantification and often do not preserve and decouple the potency and efficacy parameters of the combinations, making it difficult to understand and trust the combination candidates selected from such models. In this work, we first propose SynBa, a flexible Bayesian approach to estimate the uncertainty of the synergistic efficacy and potency of drug combinations, so that actionable decisions can be derived from the model outputs. The actionability is enabled by incorporating the Hill equation into SynBa, so that the parameters representing the potency and the efficacy can be preserved. Existing knowledge may be conveniently inserted due to the flexibility of the prior, as shown by the empirical Beta prior defined for the normalised maximal inhibition. Through experiments on large combination screenings and comparison against benchmark methods, we show that SynBa provides improved accuracy of dose-response predictions and better-calibrated uncertainty estimation for the parameters and the predictions. We then extend SynBa to SynBa-Batch to explain away the batch effects existing in the noisy measurements so that the noise can be better understood. We provide a route for SynBa and SynBa-Batch to utilise existing monotherapy information and design more informed priors. This approach enhances the reliability and robustness of inferring the synergistic efficacy and potency of drug combinations. Furthermore, we incorporate SynBa likelihood in the prediction for drug combination responses and develop DeepSynBa, a deep learning model that predicts the complete dose-response matrix of drug pairs instead of relying on an aggregated synergy score. DeepSynBa separates out efficacy and potency through the incorporation of SynBa likelihood. This enables more informed decision-making, as it offers more detailed view of drug interactions beyond a single synergy measure. Through SynBa, SynBa-Batch and DeepSynBa, we advocate the use of quantification frameworks in pharmacology that includes a principled uncertainty estimation for model parameters, and preserves and decouples the synergistic efficacy and potency of the combinations to encourage actionability of the model in further decision-making.","abstract_html":"In pharmacology, there exists a range of different quantification frameworks to estimate the synergistic effect of drug combinations. However, the varying assumptions and the absence of a universal gold standard across these frameworks make it difficult to quantitatively prioritise which drug combinations from large screening experiments should advance for further investigation. Furthermore, existing frameworks lack accurate uncertainty quantification and often do not preserve and decouple the potency and efficacy parameters of the combinations, making it difficult to understand and trust the combination candidates selected from such models. In this work, we first propose SynBa, a flexible Bayesian approach to estimate the uncertainty of the synergistic efficacy and potency of drug combinations, so that actionable decisions can be derived from the model outputs. The actionability is enabled by incorporating the Hill equation into SynBa, so that the parameters representing the potency and the efficacy can be preserved. Existing knowledge may be conveniently inserted due to the flexibility of the prior, as shown by the empirical Beta prior defined for the normalised maximal inhibition. Through experiments on large combination screenings and comparison against benchmark methods, we show that SynBa provides improved accuracy of dose-response predictions and better-calibrated uncertainty estimation for the parameters and the predictions. We then extend SynBa to SynBa-Batch to explain away the batch effects existing in the noisy measurements so that the noise can be better understood. We provide a route for SynBa and SynBa-Batch to utilise existing monotherapy information and design more informed priors. This approach enhances the reliability and robustness of inferring the synergistic efficacy and potency of drug combinations. Furthermore, we incorporate SynBa likelihood in the prediction for drug combination responses and develop DeepSynBa, a deep learning model that predicts the complete dose-response matrix of drug pairs instead of relying on an aggregated synergy score. DeepSynBa separates out efficacy and potency through the incorporation of SynBa likelihood. This enables more informed decision-making, as it offers more detailed view of drug interactions beyond a single synergy measure. Through SynBa, SynBa-Batch and DeepSynBa, we advocate the use of quantification frameworks in pharmacology that includes a principled uncertainty estimation for model parameters, and preserves and decouples the synergistic efficacy and potency of the combinations to encourage actionability of the model in further decision-making.","abstract_has_math":false,"creators":["Zhang, Haoting"],"institution":"University of Cambridge","degree_name":"Doctor of Philosophy (PhD)","degree_level":"Doctoral","degree_discipline":null,"degree_department":null,"school":null,"contributors":[],"advisors":["Ek, Carl Henrik","Milo, Marta","Rattray, Magnus"],"committee_chairs":[],"committee_members":[],"year":2024,"date_issued":"2024-09-30","date_published":"2024-09-30","updated_at":"2026-07-22T22:24:17Z","subjects":["Bayesian Inference","Drug Combination","Drug Discovery","Drug Synergy","Machine Learning"],"languages":["eng"],"rights":[],"rights_urls":["https://apollo8-f-pro.lib.cam.ac.uk/bitstreams/617c20f1-5a55-409a-826f-757576d785b6/download","https://creativecommons.org/licenses/by/4.0/"],"identifier_entries":[]},"links":{"outbound_url":"https://doi.org/10.17863/CAM.117763","outbound_label":"DOI","outbound_source":"dc:identifier.doi"},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:contributor.advisor","label":"Advisor","values":["Ek, Carl Henrik","Milo, Marta","Rattray, Magnus"]},{"key":"dc:contributor.sponsor","label":"Sponsor","values":["I acknowledge the receipt of studentship award from the Health Data Research UK-The Alan Turing Institute Wellcome PhD Programme in Health Data Science (Grant Ref: 218529/Z/19/Z) and the Wellcome Cambridge Trust Scholarship."]},{"key":"dc:creator","label":"Author","values":["Zhang, Haoting"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:date.issued","label":"Date","values":["2024-09-30"]},{"key":"dc:publisher.institution","label":"Dc Publisher Institution","values":["University of Cambridge"]},{"key":"dc:relation.isreferencedby.uri","label":"Dc Relation Isreferencedby URI","values":["https://www.repository.cam.ac.uk/handle/1810/383355"]},{"key":"dc:type","label":"Dc Type","values":["Thesis"]},{"key":"dc:type.qualificationlevel","label":"Dc Type Qualificationlevel","values":["Doctoral"]},{"key":"dc:type.qualificationname","label":"Dc Type Qualificationname","values":["Doctor of Philosophy (PhD)"]}]},{"id":"subjects_keywords","label":"Subjects and Keywords","entries":[{"key":"dc:subject","label":"Dc Subject","values":["Bayesian Inference","Drug Combination","Drug Discovery","Drug Synergy","Machine Learning"]}]},{"id":"language_rights","label":"Language and Rights","entries":[{"key":"dc:language","label":"Dc Language","values":["eng"]},{"key":"dc:rights","label":"Dc Rights","values":["https://apollo8-f-pro.lib.cam.ac.uk/bitstreams/617c20f1-5a55-409a-826f-757576d785b6/download","https://creativecommons.org/licenses/by/4.0/"]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier.doi","label":"DOI","values":["https://doi.org/10.17863/CAM.117763"]},{"key":"dc:identifier.uri","label":"Identifier URI","values":["https://apollo8-f-pro.lib.cam.ac.uk/bitstreams/2ccb87a0-b27e-4b11-bff4-7ba3cde25cc8/download"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description.abstract","label":"Abstract","values":["In pharmacology, there exists a range of different quantification frameworks to estimate the synergistic effect of drug combinations. However, the varying assumptions and the absence of a universal gold standard across these frameworks make it difficult to quantitatively prioritise which drug combinations from large screening experiments should advance for further investigation. Furthermore, existing frameworks lack accurate uncertainty quantification and often do not preserve and decouple the potency and efficacy parameters of the combinations, making it difficult to understand and trust the combination candidates selected from such models. In this work, we first propose SynBa, a flexible Bayesian approach to estimate the uncertainty of the synergistic efficacy and potency of drug combinations, so that actionable decisions can be derived from the model outputs. The actionability is enabled by incorporating the Hill equation into SynBa, so that the parameters representing the potency and the efficacy can be preserved. Existing knowledge may be conveniently inserted due to the flexibility of the prior, as shown by the empirical Beta prior defined for the normalised maximal inhibition. Through experiments on large combination screenings and comparison against benchmark methods, we show that SynBa provides improved accuracy of dose-response predictions and better-calibrated uncertainty estimation for the parameters and the predictions. We then extend SynBa to SynBa-Batch to explain away the batch effects existing in the noisy measurements so that the noise can be better understood. We provide a route for SynBa and SynBa-Batch to utilise existing monotherapy information and design more informed priors. This approach enhances the reliability and robustness of inferring the synergistic efficacy and potency of drug combinations. Furthermore, we incorporate SynBa likelihood in the prediction for drug combination responses and develop DeepSynBa, a deep learning model that predicts the complete dose-response matrix of drug pairs instead of relying on an aggregated synergy score. DeepSynBa separates out efficacy and potency through the incorporation of SynBa likelihood. This enables more informed decision-making, as it offers more detailed view of drug interactions beyond a single synergy measure. Through SynBa, SynBa-Batch and DeepSynBa, we advocate the use of quantification frameworks in pharmacology that includes a principled uncertainty estimation for model parameters, and preserves and decouples the synergistic efficacy and potency of the combinations to encourage actionability of the model in further decision-making."]},{"key":"dc:format.checksum.md5","label":"Dc Format Checksum Md5","values":["c37a6f02803434a4d5c7320a95986ad0","87eda9de84448d1f82354d60eee3eb5f"]},{"key":"dc:title","label":"Title","values":["Accurate estimation of drug combination synergies with uncertainty quantification"]}]}],"canonical_facts":{"dc:contributor.advisor":["Ek, Carl Henrik","Milo, Marta","Rattray, Magnus"],"dc:contributor.sponsor":["I acknowledge the receipt of studentship award from the Health Data Research UK-The Alan Turing Institute Wellcome PhD Programme in Health Data Science (Grant Ref: 218529/Z/19/Z) and the Wellcome Cambridge Trust Scholarship."],"dc:creator":["Zhang, Haoting"],"dc:date.issued":["2024-09-30"],"dc:description.abstract":["In pharmacology, there exists a range of different quantification frameworks to estimate the synergistic effect of drug combinations. However, the varying assumptions and the absence of a universal gold standard across these frameworks make it difficult to quantitatively prioritise which drug combinations from large screening experiments should advance for further investigation. Furthermore, existing frameworks lack accurate uncertainty quantification and often do not preserve and decouple the potency and efficacy parameters of the combinations, making it difficult to understand and trust the combination candidates selected from such models. In this work, we first propose SynBa, a flexible Bayesian approach to estimate the uncertainty of the synergistic efficacy and potency of drug combinations, so that actionable decisions can be derived from the model outputs. The actionability is enabled by incorporating the Hill equation into SynBa, so that the parameters representing the potency and the efficacy can be preserved. Existing knowledge may be conveniently inserted due to the flexibility of the prior, as shown by the empirical Beta prior defined for the normalised maximal inhibition. Through experiments on large combination screenings and comparison against benchmark methods, we show that SynBa provides improved accuracy of dose-response predictions and better-calibrated uncertainty estimation for the parameters and the predictions. We then extend SynBa to SynBa-Batch to explain away the batch effects existing in the noisy measurements so that the noise can be better understood. We provide a route for SynBa and SynBa-Batch to utilise existing monotherapy information and design more informed priors. This approach enhances the reliability and robustness of inferring the synergistic efficacy and potency of drug combinations. Furthermore, we incorporate SynBa likelihood in the prediction for drug combination responses and develop DeepSynBa, a deep learning model that predicts the complete dose-response matrix of drug pairs instead of relying on an aggregated synergy score. DeepSynBa separates out efficacy and potency through the incorporation of SynBa likelihood. This enables more informed decision-making, as it offers more detailed view of drug interactions beyond a single synergy measure. Through SynBa, SynBa-Batch and DeepSynBa, we advocate the use of quantification frameworks in pharmacology that includes a principled uncertainty estimation for model parameters, and preserves and decouples the synergistic efficacy and potency of the combinations to encourage actionability of the model in further decision-making."],"dc:format.checksum.md5":["c37a6f02803434a4d5c7320a95986ad0","87eda9de84448d1f82354d60eee3eb5f"],"dc:identifier.doi":["https://doi.org/10.17863/CAM.117763"],"dc:identifier.uri":["https://apollo8-f-pro.lib.cam.ac.uk/bitstreams/2ccb87a0-b27e-4b11-bff4-7ba3cde25cc8/download"],"dc:language":["eng"],"dc:publisher.institution":["University of Cambridge"],"dc:relation.isreferencedby.uri":["https://www.repository.cam.ac.uk/handle/1810/383355"],"dc:rights":["https://apollo8-f-pro.lib.cam.ac.uk/bitstreams/617c20f1-5a55-409a-826f-757576d785b6/download","https://creativecommons.org/licenses/by/4.0/"],"dc:subject":["Bayesian Inference","Drug Combination","Drug Discovery","Drug Synergy","Machine Learning"],"dc:title":["Accurate estimation of drug combination synergies with uncertainty quantification"],"dc:type":["Thesis"],"dc:type.qualificationlevel":["Doctoral"],"dc:type.qualificationname":["Doctor of Philosophy (PhD)"]},"updated_at":"2026-07-22T22:24:17Z"}