{"id":{"repo_id":"missouri","oai_identifier":"oai:mospace.umsystem.edu:10355/111003"},"canonical_url":"https://search.dev.ndltd.org/etd/missouri/oai:mospace.umsystem.edu:10355/111003","repository":{"repo_id":"missouri","name":"University of Missouri","base_url":"https://mospace.umsystem.edu/oai/request"},"display":{"title":"ScPEFT : a parameter-efficient fine-tuning framework for enhancing single-cell large language models in out-of-context","abstract":"Single-cell large language models (scLLMs) capture essential biological insights from vast single-cell atlases but struggle in out-of-context applications, where zero-shot predictions can be unreliable. To address this, we introduce a single-cell parameter-efficient fine-tuning (scPEFT) framework that integrates learnable, low-dimensional adapters into scLLMs. By freezing the backbone model and updating only the adapter parameters, scPEFT efficiently adapts to specific tasks using limited custom data. This approach mitigates catastrophic forgetting, reduces parameter tuning by over 96%, and decreases GPU memory usage by more than half, significantly enhancing scLLMs's accessibility for resourceconstrained researchers. Validated across diverse datasets, scPEFT outperformed zero-shot models and traditional finetuning in disease-specific, cross-species, and under-characterized cell population tasks. Its attention-mechanism analysis identified COVID-related genes associated with specific cell states and uncovered unique blood cell subpopulations, demonstrating scPEFT's capacity for conditionspecific interpretations. These findings position scPEFT as an efficient solution for improving scLLMs' utilities in general single-cell analyses.","abstract_html":"Single-cell large language models (scLLMs) capture essential biological insights from vast single-cell atlases but struggle in out-of-context applications, where zero-shot predictions can be unreliable. To address this, we introduce a single-cell parameter-efficient fine-tuning (scPEFT) framework that integrates learnable, low-dimensional adapters into scLLMs. By freezing the backbone model and updating only the adapter parameters, scPEFT efficiently adapts to specific tasks using limited custom data. This approach mitigates catastrophic forgetting, reduces parameter tuning by over 96%, and decreases GPU memory usage by more than half, significantly enhancing scLLMs's accessibility for resourceconstrained researchers. Validated across diverse datasets, scPEFT outperformed zero-shot models and traditional finetuning in disease-specific, cross-species, and under-characterized cell population tasks. Its attention-mechanism analysis identified COVID-related genes associated with specific cell states and uncovered unique blood cell subpopulations, demonstrating scPEFT's capacity for conditionspecific interpretations. These findings position scPEFT as an efficient solution for improving scLLMs' utilities in general single-cell analyses.","abstract_has_math":false,"creators":["He, Fei"],"institution":"University of Missouri--Columbia","degree_name":"Ph. D.","degree_level":"Doctoral","degree_discipline":"Computer Science","degree_department":null,"school":null,"contributors":[],"advisors":["Xu, Dong"],"committee_chairs":[],"committee_members":[],"year":2025,"date_issued":"2025","date_published":"2025","updated_at":"2026-07-24T03:07:13Z","subjects":[],"languages":["eng","English"],"rights":[],"rights_urls":[],"identifier_entries":[{"key":"dc:identifier.doi","label":"DOI","values":["https://doi.org/10.32469/10355/111003"],"render_values":[{"text":"https://doi.org/10.32469/10355/111003","href":"https://doi.org/10.32469/10355/111003","code":true}]}]},"links":{"outbound_url":"https://hdl.handle.net/10355/111003","outbound_label":"Handle","outbound_source":"dc:identifier.uri"},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:contributor.advisor","label":"Advisor","values":["Xu, Dong"]},{"key":"dc:creator","label":"Author","values":["He, Fei"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:date.accessioned","label":"Dc Date Accessioned","values":["2026-03-02T17:22:01Z"]},{"key":"dc:date.available","label":"Dc Date Available","values":["2026-03-02T17:22:01Z"]},{"key":"dc:date.issued","label":"Date","values":["2025"]},{"key":"dc:publisher","label":"Institution","values":["University of Missouri--Columbia"]},{"key":"dc:type","label":"Dc Type","values":["Thesis"]},{"key":"thesis:degree_discipline","label":"Discipline","values":["Computer Science"]},{"key":"thesis:degree_level","label":"Degree Level","values":["Doctoral"]},{"key":"thesis:degree_name","label":"Degree Name","values":["Ph. D."]},{"key":"thesis:institution_name","label":"Thesis Institution Name","values":["University of Missouri--Columbia"]}]},{"id":"language_rights","label":"Language and Rights","entries":[{"key":"dc:language","label":"Dc Language","values":["English"]},{"key":"dc:language.iso","label":"Language (ISO)","values":["eng"]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier.doi","label":"DOI","values":["https://doi.org/10.32469/10355/111003"]},{"key":"dc:identifier.uri","label":"Identifier URI","values":["https://hdl.handle.net/10355/111003"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description.abstract","label":"Abstract","values":["Single-cell large language models (scLLMs) capture essential biological insights from vast single-cell atlases but struggle in out-of-context applications, where zero-shot predictions can be unreliable. To address this, we introduce a single-cell parameter-efficient fine-tuning (scPEFT) framework that integrates learnable, low-dimensional adapters into scLLMs. By freezing the backbone model and updating only the adapter parameters, scPEFT efficiently adapts to specific tasks using limited custom data. This approach mitigates catastrophic forgetting, reduces parameter tuning by over 96%, and decreases GPU memory usage by more than half, significantly enhancing scLLMs's accessibility for resourceconstrained researchers. Validated across diverse datasets, scPEFT outperformed zero-shot models and traditional finetuning in disease-specific, cross-species, and under-characterized cell population tasks. Its attention-mechanism analysis identified COVID-related genes associated with specific cell states and uncovered unique blood cell subpopulations, demonstrating scPEFT's capacity for conditionspecific interpretations. These findings position scPEFT as an efficient solution for improving scLLMs' utilities in general single-cell analyses."]},{"key":"dc:title","label":"Title","values":["ScPEFT : a parameter-efficient fine-tuning framework for enhancing single-cell large language models in out-of-context"]}]}],"canonical_facts":{"dc:contributor.advisor":["Xu, Dong"],"dc:creator":["He, Fei"],"dc:date.accessioned":["2026-03-02T17:22:01Z"],"dc:date.available":["2026-03-02T17:22:01Z"],"dc:date.issued":["2025"],"dc:description.abstract":["Single-cell large language models (scLLMs) capture essential biological insights from vast single-cell atlases but struggle in out-of-context applications, where zero-shot predictions can be unreliable. To address this, we introduce a single-cell parameter-efficient fine-tuning (scPEFT) framework that integrates learnable, low-dimensional adapters into scLLMs. By freezing the backbone model and updating only the adapter parameters, scPEFT efficiently adapts to specific tasks using limited custom data. This approach mitigates catastrophic forgetting, reduces parameter tuning by over 96%, and decreases GPU memory usage by more than half, significantly enhancing scLLMs's accessibility for resourceconstrained researchers. Validated across diverse datasets, scPEFT outperformed zero-shot models and traditional finetuning in disease-specific, cross-species, and under-characterized cell population tasks. Its attention-mechanism analysis identified COVID-related genes associated with specific cell states and uncovered unique blood cell subpopulations, demonstrating scPEFT's capacity for conditionspecific interpretations. These findings position scPEFT as an efficient solution for improving scLLMs' utilities in general single-cell analyses."],"dc:identifier.doi":["https://doi.org/10.32469/10355/111003"],"dc:identifier.uri":["https://hdl.handle.net/10355/111003"],"dc:language":["English"],"dc:language.iso":["eng"],"dc:publisher":["University of Missouri--Columbia"],"dc:title":["ScPEFT : a parameter-efficient fine-tuning framework for enhancing single-cell large language models in out-of-context"],"dc:type":["Thesis"],"thesis:degree_discipline":["Computer Science"],"thesis:degree_level":["Doctoral"],"thesis:degree_name":["Ph. D."],"thesis:institution_name":["University of Missouri--Columbia"]},"updated_at":"2026-07-24T03:07:13Z"}