본문

서브메뉴

Towards a Systematic Understanding of Human Cells Through Data-Driven Models of Spatial Proteomics
Towards a Systematic Understanding of Human Cells Through Data-Driven Models of Spatial Pr...
Towards a Systematic Understanding of Human Cells Through Data-Driven Models of Spatial Proteomics

Detailed Information

자료유형  
 학위논문 서양
최종처리일시  
20260202105618
ISBN  
9798265427663
DDC  
600
저자명  
Le Candidate, Trang.
서명/저자  
Towards a Systematic Understanding of Human Cells Through Data-Driven Models of Spatial Proteomics
발행사항  
[Sl] : Stanford University, 2025
발행사항  
Ann Arbor : ProQuest Dissertations & Theses, 2025
형태사항  
221 p
주기사항  
Source: Dissertations Abstracts International, Volume: 87-05, Section: B.
주기사항  
Advisor: Lundberg, Emma.
학위논문주기  
Thesis (Ph.D.)--Stanford University, 2025.
초록/해제  
요약Cells are spatially organized multi-scaled systems where precise protein localization underpins cellular identity, function, and dynamic response to perturbations. Spatial proteomics provides a powerful framework to systematically characterize this organization and elucidate underlying biological mechanisms at subcellular resolution.This thesis is a body of di↵erent works that investigate cell biology from a spatial proteomic angle and provide some tools to approach modeling subcellular proteins for systematic understanding.Chapter 2 described the first e↵ort to predict single cell protein localization from confocal microscopy images. We designed and analyzed a citizen science competition to predict single cell classification labels for both single- and multilocalization. The main challenges for this representation learning competition include class imbalance, weak labels, and multi-label classification across 19 classes in 17 di↵erent human cell lines. Diverse strategies were employed, and the winning models serve as one of the first subcellular omics tools capable of capturing subcellular features, cellular dynamics and accurate single-cell labels.Chapter 3 described a signal processing and machine learning framework to create a common shapespace and map proteins representation in a common coordinate of the cell based on cell and nuclei shapes. We performed an integrated analysis linking organelle, pathway, and single-protein levels to shapespace. While subcellular organelle topology remained robust across shape spacebut varied between cell lines, single-protein analysis showed that shape differences within the same cell cycle phase may indicate distinct cell fates, with many non-cell cycle proteins exhibiting shape-dependent variation, enabling the study of protein spatial shifts under perturbation within a common coordinate framework.Chapter 4 investigated landscape of protein reorganization in host-virus under SARS-CoV-2 infection. We performed a large-scale immunofluorescence screening of 602 host proteins and identified ˜100 proteins with altered abundance or localization upon infection. Many of this proteins were validated in another cell lines, with some di↵erences highlighting the heterogeneity in response to virus infection. Using the identified di↵erential protein list, we screened 12 approved small-molecules drug targeting these proteins and identified elesclomol and rimcazole as potential antiviral candidate. Our study can serve as a systematic approach for drug repurposing, and the generated dataset of more than 100,000 immunofluorescence images was published as a resource available for further studies.Chapter 5 focused on tool development, aiming to open-source some key bioimage analysis and modeling tasks.Chapter 5.1. presented an interactive training and annotation framework, accessible via a web app for real-time monitoring and annotation, while a connected backend server handles model training. This addressed the growing role of deep learning in bioimage analysis, where the lack of user-friendly tools hindered broader adoption. Many existing platforms struggled with model re-training and fine-tuning, which are essential for preventing overfitting, particularly with limited training data. Our approach leveraged interactive model training, allowing real-time refinement, as demonstrated with segmentation models, achieving a 6x faster workflow compared to conventional methods.Chapter 5.2 described generative models for organelle staining, one steptoward the whole proteome virtual cell painting. Label-free organelle prediction, or virual cell painting, is a longstanding challenge in cellular imaging, o↵ering a way to overcome the limitations of fluorescence microscopy, such as high costs, cytotoxicity, and labor-intensive protocols. This paper propose a simplified VQGAN adaptable to various input/output channel numbers for image-to-image translation, enabling multi-channel organelle staining prediction from transmitted light images. The approach won first place in ISBI 2024 Light My Cell competition on Grand Challenge platform.
일반주제명  
Infections
일반주제명  
Software
일반주제명  
Endoplasmic reticulum
일반주제명  
Deep learning
일반주제명  
Fourier transforms
일반주제명  
Severe acute respiratory syndrome coronavirus 2
일반주제명  
Proteomics
일반주제명  
Cell cycle
일반주제명  
Bioinformatics
일반주제명  
Cellular biology
일반주제명  
Mathematics
일반주제명  
Virology
기타저자  
Stanford University.
기본자료저록  
Dissertations Abstracts International. 87-05B.
전자적 위치 및 접속  
로그인 후 원문을 볼 수 있습니다.

MARC

 008260126s2025        us                              c    eng  d
■001000017360776
■00520260202105618
■006m          o    d                
■007cr#unu||||||||
■020    ▼a9798265427663
■035    ▼a(MiAaPQ)AAI32316476
■035    ▼a(MiAaPQ)Stanfordvy798hn0705
■040    ▼aMiAaPQ▼cMiAaPQ
■0820  ▼a600
■1001  ▼aLe  Candidate,  Trang.
■24510▼aTowards  a  Systematic  Understanding  of  Human  Cells  Through  Data-Driven  Models  of  Spatial  Proteomics
■260    ▼a[Sl]▼bStanford  University▼c2025
■260  1▼aAnn  Arbor▼bProQuest  Dissertations  &  Theses▼c2025
■300    ▼a221  p
■500    ▼aSource:  Dissertations  Abstracts  International,  Volume:  87-05,  Section:  B.
■500    ▼aAdvisor:  Lundberg,  Emma.
■5021  ▼aThesis  (Ph.D.)--Stanford  University,  2025.
■520    ▼aCells  are  spatially  organized  multi-scaled  systems  where  precise  protein  localization  underpins  cellular  identity,  function,  and  dynamic  response  to  perturbations.  Spatial  proteomics  provides  a  powerful  framework  to  systematically  characterize  this  organization  and  elucidate  underlying  biological  mechanisms  at  subcellular  resolution.This  thesis  is  a  body  of  di↵erent  works  that  investigate  cell  biology  from  a  spatial  proteomic  angle  and  provide  some  tools  to  approach  modeling  subcellular  proteins  for  systematic  understanding.Chapter  2  described  the  first  e↵ort  to  predict  single  cell  protein  localization  from  confocal  microscopy  images.  We  designed  and  analyzed  a  citizen  science  competition  to  predict  single  cell  classification  labels  for  both  single-  and  multilocalization.  The  main  challenges  for  this  representation  learning  competition  include  class  imbalance,  weak  labels,  and  multi-label  classification  across  19  classes  in  17  di↵erent  human  cell  lines.  Diverse  strategies  were  employed,  and  the  winning  models  serve  as  one  of  the  first  subcellular  omics  tools  capable  of  capturing  subcellular  features,  cellular  dynamics  and  accurate  single-cell  labels.Chapter  3  described  a  signal  processing  and  machine  learning  framework  to  create  a  common  shapespace  and  map  proteins  representation  in  a  common  coordinate  of  the  cell  based  on  cell  and  nuclei  shapes.  We  performed  an  integrated  analysis  linking  organelle,  pathway,  and  single-protein  levels  to  shapespace.  While  subcellular  organelle  topology  remained  robust  across  shape  spacebut  varied  between  cell  lines,  single-protein  analysis  showed  that  shape  differences  within  the  same  cell  cycle  phase  may  indicate  distinct  cell  fates,  with  many  non-cell  cycle  proteins  exhibiting  shape-dependent  variation,  enabling  the  study  of  protein  spatial  shifts  under  perturbation  within  a  common  coordinate  framework.Chapter  4  investigated  landscape  of  protein  reorganization  in  host-virus  under  SARS-CoV-2  infection.  We  performed  a  large-scale  immunofluorescence  screening  of  602  host  proteins  and  identified  ˜100  proteins  with  altered  abundance  or  localization  upon  infection.  Many  of  this  proteins  were  validated  in  another  cell  lines,  with  some  di↵erences  highlighting  the  heterogeneity  in  response  to  virus  infection.  Using  the  identified  di↵erential  protein  list,  we  screened  12  approved  small-molecules  drug  targeting  these  proteins  and  identified  elesclomol  and  rimcazole  as  potential  antiviral  candidate.  Our  study  can  serve  as  a  systematic  approach  for  drug  repurposing,  and  the  generated  dataset  of  more  than  100,000  immunofluorescence  images  was  published  as  a  resource  available  for  further  studies.Chapter  5  focused  on  tool  development,  aiming  to  open-source  some  key  bioimage  analysis  and  modeling  tasks.Chapter  5.1.  presented  an  interactive  training  and  annotation  framework,  accessible  via  a  web  app  for  real-time  monitoring  and  annotation,  while  a  connected  backend  server  handles  model  training.  This  addressed  the  growing  role  of  deep  learning  in  bioimage  analysis,  where  the  lack  of  user-friendly  tools  hindered  broader  adoption.  Many  existing  platforms  struggled  with  model  re-training  and  fine-tuning,  which  are  essential  for  preventing  overfitting,  particularly  with  limited  training  data.  Our  approach  leveraged  interactive  model  training,  allowing  real-time  refinement,  as  demonstrated  with  segmentation  models,  achieving  a  6x  faster  workflow  compared  to  conventional  methods.Chapter  5.2  described  generative  models  for  organelle  staining,  one  steptoward  the  whole  proteome  virtual  cell  painting.  Label-free  organelle  prediction,  or  virual  cell  painting,  is  a  longstanding  challenge  in  cellular  imaging,  o↵ering  a  way  to  overcome  the  limitations  of  fluorescence  microscopy,  such  as  high  costs,  cytotoxicity,  and  labor-intensive  protocols.  This  paper  propose  a  simplified  VQGAN  adaptable  to  various  input/output  channel  numbers  for  image-to-image  translation,  enabling  multi-channel  organelle  staining  prediction  from  transmitted  light  images.  The  approach  won  first  place  in  ISBI  2024  Light  My  Cell  competition  on  Grand  Challenge  platform.
■590    ▼aSchool  code:  0212.
■650  4▼aInfections
■650  4▼aSoftware
■650  4▼aEndoplasmic  reticulum
■650  4▼aDeep  learning
■650  4▼aFourier  transforms
■650  4▼aSevere  acute  respiratory  syndrome  coronavirus  2
■650  4▼aProteomics
■650  4▼aCell  cycle
■650  4▼aBioinformatics
■650  4▼aCellular  biology
■650  4▼aMathematics
■650  4▼aVirology
■690    ▼a0800
■690    ▼a0715
■690    ▼a0379
■690    ▼a0405
■690    ▼a0720
■71020▼aStanford  University.
■7730  ▼tDissertations  Abstracts  International▼g87-05B.
■790    ▼a0212
■791    ▼aPh.D.
■792    ▼a2025
■793    ▼aEnglish
■85640▼uhttp://www.riss.kr/pdu/ddodLink.do?id=T17360776▼nKERIS▼z이  자료의  원문은  한국교육학술정보원에서  제공합니다.

Preview

Export

ChatGPT Discussion

AI Recommended Related Books


    New Books MORE
    Statistics for the past 3 years. Go to brief

    Подробнее информация.

    • Бронирование
    • не существует
    • моя папка
    • Первый запрос зрения
    • Non-Book Loan Application
    • Nighttime Book Loan Application
    материал
    Reg No. Количество платежных Местоположение статус Ленд информации
    TF15522 전자도서 대출가능 My Folder 부재도서신고 비도서대출신청 야간 도서대출신청

    * Бронирование доступны в заимствований книги. Чтобы сделать предварительный заказ, пожалуйста, нажмите кнопку бронирование

    Books borrowed together with this book

    Related Popular Books

    Available after logging in.