서브메뉴
검색
Computational Methods for Linking Molecular and Anatomical Data With Spatial Transcriptomics
Computational Methods for Linking Molecular and Anatomical Data With Spatial Transcriptomics
Detailed Information
- 자료유형
- 학위논문 서양
- 최종처리일시
- 20250211153012
- ISBN
- 9798384044987
- DDC
- 574
- 서명/저자
- Computational Methods for Linking Molecular and Anatomical Data With Spatial Transcriptomics
- 발행사항
- [Sl] : University of Michigan, 2024
- 발행사항
- Ann Arbor : ProQuest Dissertations & Theses, 2024
- 형태사항
- 222 p
- 주기사항
- Source: Dissertations Abstracts International, Volume: 86-03, Section: B.
- 주기사항
- Advisor: Welch, Joshua.
- 학위논문주기
- Thesis (Ph.D.)--University of Michigan, 2024.
- 초록/해제
- 요약The ideology of, "To understand the whole, we must first understand its parts" lies behind much of recent scientific progress, including the Human Cell Atlas Project. To better understand the body, the Human Cell Atlas Project strives to define what cells compose our bodies. A cell's identity is complex, with some defining characteristics including a cell's transcriptome, epigenome, spatial location within an organ and in relation to other cells, and how that cell functions from a physiological perspective. While most technologies typically assess one or two of these aspects of cellular identity, we cannot yet measure all facets in a single assay. Consequently, existing computational methods must focus on stitching these individual fragments of cellular identity into a single, cohesive cellular profile. The focus of this dissertation is the development of computational tools to facilitate the integration of molecular, spatial, anatomical, and physiological data to define more panoramic cellular profiles.In Chapter II, we develop UINMF, an integrative non-negative matrix factorization algorithm. Most methods are limited to integrating single-cell datasets using only the features that are shared between all datasets in the integration. Leveraging only shared features within an integration is particularly problematic for spatial transcriptomic datasets with few genes and cross-species integrations that are restricted to the set of one-to-one orthologous genes between species. UINMF accommodates the unshared features into each iteration of the objective function, allowing relevant unshared features to help shape the latent space. We show that the use of UINMF to include unshared features improves cross-modality and cross-species analyses.In Chapter III, we generate a whole-brain molecular mouse atlas by simultaneously integrating six distinct modalities. We integrate epigenomic and transcriptomic single-cell and single-nucleus datasets in 18 regions across the mouse brain, with each region being analyzed in 3 refined sub-analyses: non-neuronal cells, excitatory neurons, and inhibitory and other neurons. After manually annotating each cluster, we derive probable spatial distributions for the molecular cell type profiles by deconvolving two spatial transcriptomics datasets with two separate deconvolution algorithms: an established algorithm, RCTD, and a novel deconvolution algorithm, SiNMFiD. We created SiNMFiD for the unique challenge of deconvolving expression data with low resolution. Consequently, we construct a whole-brain mouse atlas with jointly defined molecular profiles and accompanying spatial distributions.In Chapter IV, we leverage a common coordinate framework to integrate the spatially resolved molecular profiles from our whole-brain atlas with anatomical and physiological datasets. Specifically, we explore the correlation between molecular cell types and vascular density, neuronal projection data, and c-Fos activity after foot shock. We determine which molecular cell type profiles are highly and lowly correlated with vascular density and identify distinct cell types that correspond with specific neuronal projection patterns. Using c-Fos data captured after foot shock, we capture the molecular cell types most associated with increased c-Fos activity over time after a foot shock experiment.
- 일반주제명
- Bioinformatics
- 일반주제명
- Cellular biology
- 일반주제명
- Molecular biology
- 일반주제명
- Physiology
- 키워드
- Single cell
- 기타저자
- University of Michigan Bioinformatics
- 기본자료저록
- Dissertations Abstracts International. 86-03B.
- 전자적 위치 및 접속
- 로그인 후 원문을 볼 수 있습니다.
MARC
008250123s2024 us c eng d■001000017164514
■00520250211153012
■006m o d
■007cr#unu||||||||
■020 ▼a9798384044987
■035 ▼a(MiAaPQ)AAI31631463
■035 ▼a(MiAaPQ)umichrackham005584
■040 ▼aMiAaPQ▼cMiAaPQ
■0820 ▼a574
■1001 ▼aKriebel, April R.
■24510▼aComputational Methods for Linking Molecular and Anatomical Data With Spatial Transcriptomics
■260 ▼a[Sl]▼bUniversity of Michigan▼c2024
■260 1▼aAnn Arbor▼bProQuest Dissertations & Theses▼c2024
■300 ▼a222 p
■500 ▼aSource: Dissertations Abstracts International, Volume: 86-03, Section: B.
■500 ▼aAdvisor: Welch, Joshua.
■5021 ▼aThesis (Ph.D.)--University of Michigan, 2024.
■520 ▼aThe ideology of, "To understand the whole, we must first understand its parts" lies behind much of recent scientific progress, including the Human Cell Atlas Project. To better understand the body, the Human Cell Atlas Project strives to define what cells compose our bodies. A cell's identity is complex, with some defining characteristics including a cell's transcriptome, epigenome, spatial location within an organ and in relation to other cells, and how that cell functions from a physiological perspective. While most technologies typically assess one or two of these aspects of cellular identity, we cannot yet measure all facets in a single assay. Consequently, existing computational methods must focus on stitching these individual fragments of cellular identity into a single, cohesive cellular profile. The focus of this dissertation is the development of computational tools to facilitate the integration of molecular, spatial, anatomical, and physiological data to define more panoramic cellular profiles.In Chapter II, we develop UINMF, an integrative non-negative matrix factorization algorithm. Most methods are limited to integrating single-cell datasets using only the features that are shared between all datasets in the integration. Leveraging only shared features within an integration is particularly problematic for spatial transcriptomic datasets with few genes and cross-species integrations that are restricted to the set of one-to-one orthologous genes between species. UINMF accommodates the unshared features into each iteration of the objective function, allowing relevant unshared features to help shape the latent space. We show that the use of UINMF to include unshared features improves cross-modality and cross-species analyses.In Chapter III, we generate a whole-brain molecular mouse atlas by simultaneously integrating six distinct modalities. We integrate epigenomic and transcriptomic single-cell and single-nucleus datasets in 18 regions across the mouse brain, with each region being analyzed in 3 refined sub-analyses: non-neuronal cells, excitatory neurons, and inhibitory and other neurons. After manually annotating each cluster, we derive probable spatial distributions for the molecular cell type profiles by deconvolving two spatial transcriptomics datasets with two separate deconvolution algorithms: an established algorithm, RCTD, and a novel deconvolution algorithm, SiNMFiD. We created SiNMFiD for the unique challenge of deconvolving expression data with low resolution. Consequently, we construct a whole-brain mouse atlas with jointly defined molecular profiles and accompanying spatial distributions.In Chapter IV, we leverage a common coordinate framework to integrate the spatially resolved molecular profiles from our whole-brain atlas with anatomical and physiological datasets. Specifically, we explore the correlation between molecular cell types and vascular density, neuronal projection data, and c-Fos activity after foot shock. We determine which molecular cell type profiles are highly and lowly correlated with vascular density and identify distinct cell types that correspond with specific neuronal projection patterns. Using c-Fos data captured after foot shock, we capture the molecular cell types most associated with increased c-Fos activity over time after a foot shock experiment.
■590 ▼aSchool code: 0127.
■650 4▼aBioinformatics
■650 4▼aCellular biology
■650 4▼aMolecular biology
■650 4▼aPhysiology
■653 ▼aSingle cell
■653 ▼aIntegrating molecular data
■653 ▼aSingle cell brain atlas
■653 ▼aSpatial transcriptomic
■690 ▼a0715
■690 ▼a0379
■690 ▼a0307
■690 ▼a0719
■71020▼aUniversity of Michigan▼bBioinformatics.
■7730 ▼tDissertations Abstracts International▼g86-03B.
■790 ▼a0127
■791 ▼aPh.D.
■792 ▼a2024
■793 ▼aEnglish
■85640▼uhttp://www.riss.kr/pdu/ddodLink.do?id=T17164514▼nKERIS▼z이 자료의 원문은 한국교육학술정보원에서 제공합니다.
Preview
Export
ChatGPT Discussion
AI Recommended Related Books
Подробнее информация.
- Бронирование
- не существует
- моя папка
- Первый запрос зрения
- Non-Book Loan Application
- Nighttime Book Loan Application
Available after logging in.


