서브메뉴
검색
From Structural Clustering to Enhanced Sampling: A Data-Driven Approach for Exploring Protein Conformational Ensembles
From Structural Clustering to Enhanced Sampling: A Data-Driven Approach for Exploring Protein Conformational Ensembles
상세정보
- 자료유형
- 학위논문 서양
- 최종처리일시
- 20260202103048
- ISBN
- 9798286425440
- DDC
- 542
- 저자명
- Sasmal, Subarna.
- 서명/저자
- From Structural Clustering to Enhanced Sampling: A Data-Driven Approach for Exploring Protein Conformational Ensembles
- 발행사항
- [Sl] : New York University, 2025
- 발행사항
- Ann Arbor : ProQuest Dissertations & Theses, 2025
- 형태사항
- 147 p
- 주기사항
- Source: Dissertations Abstracts International, Volume: 86-12, Section: B.
- 주기사항
- Advisor: Hocky, Glen M.
- 학위논문주기
- Thesis (Ph.D.)--New York University, 2025.
- 초록/해제
- 요약Proteins are a class of biomolecules that are one of the most important building blocks of living organisms. While it is common knowledge that the function of a protein is determined by its three dimensional structure, in reality proteins exist in multiple metastable states with different energy and specific functions. Molecular dynamics simulations is an approach by which we can use computational modeling to characterize the proteins conformational ensemble with atomistic detail.In practice, simple simulations do not allow us to access relevant conformations in accessible amounts of conformational time.Enhanced sampling algorithms allow us to more rapidly explore a system's conformational ensemble, but typically requires guessing a set of collective coordinates that, if biased, would allow us to observe all relevant configurations with an inference of their correct likelihoods. In this work, I describe an approach for simultaneously characterizing and exploring conformational ensembles of proteins.Our approach relies on a probabilistic clustering model called ShapeGMM, where configurations are used to learn a Gaussian mixture model in cartesian coordinate space. In my work, we demonstrated that the technique Linear Discriminant Analysis can be used to form a coordinate that separates states of a molecule and allows us to sample between them using enhanced sampling. We then showed that we can train a ShapeGMM model with samples generated by such a bias approach. This gives an approach by which conformational ensembles can be quantitatively characterized. Finally, we show that this allows us to perform iteration, in which case we can develop better coordinates by alternating sampling and fitting.
- 일반주제명
- Computational chemistry
- 일반주제명
- Chemistry
- 일반주제명
- Biophysics
- 일반주제명
- Molecular chemistry
- 기타저자
- New York University Chemistry
- 기본자료저록
- Dissertations Abstracts International. 86-12B.
- 전자적 위치 및 접속
- 로그인 후 원문을 볼 수 있습니다.
MARC
008260126s2025 us c eng d■001000017356848
■00520260202103048
■006m o d
■007cr#unu||||||||
■020 ▼a9798286425440
■035 ▼a(MiAaPQ)AAI31931157
■040 ▼aMiAaPQ▼cMiAaPQ
■0820 ▼a542
■1001 ▼aSasmal, Subarna.
■24510▼aFrom Structural Clustering to Enhanced Sampling: A Data-Driven Approach for Exploring Protein Conformational Ensembles
■260 ▼a[Sl]▼bNew York University▼c2025
■260 1▼aAnn Arbor▼bProQuest Dissertations & Theses▼c2025
■300 ▼a147 p
■500 ▼aSource: Dissertations Abstracts International, Volume: 86-12, Section: B.
■500 ▼aAdvisor: Hocky, Glen M.
■5021 ▼aThesis (Ph.D.)--New York University, 2025.
■520 ▼aProteins are a class of biomolecules that are one of the most important building blocks of living organisms. While it is common knowledge that the function of a protein is determined by its three dimensional structure, in reality proteins exist in multiple metastable states with different energy and specific functions. Molecular dynamics simulations is an approach by which we can use computational modeling to characterize the proteins conformational ensemble with atomistic detail.In practice, simple simulations do not allow us to access relevant conformations in accessible amounts of conformational time.Enhanced sampling algorithms allow us to more rapidly explore a system's conformational ensemble, but typically requires guessing a set of collective coordinates that, if biased, would allow us to observe all relevant configurations with an inference of their correct likelihoods. In this work, I describe an approach for simultaneously characterizing and exploring conformational ensembles of proteins.Our approach relies on a probabilistic clustering model called ShapeGMM, where configurations are used to learn a Gaussian mixture model in cartesian coordinate space. In my work, we demonstrated that the technique Linear Discriminant Analysis can be used to form a coordinate that separates states of a molecule and allows us to sample between them using enhanced sampling. We then showed that we can train a ShapeGMM model with samples generated by such a bias approach. This gives an approach by which conformational ensembles can be quantitatively characterized. Finally, we show that this allows us to perform iteration, in which case we can develop better coordinates by alternating sampling and fitting.
■590 ▼aSchool code: 0146.
■650 4▼aComputational chemistry
■650 4▼aChemistry
■650 4▼aBiophysics
■650 4▼aMolecular chemistry
■653 ▼aClustering algorithms
■653 ▼aDimensionality reduction
■653 ▼aEnhanced sampling
■653 ▼aFree energy calculation
■653 ▼aMolecular dynamics
■653 ▼aMolecular modeling
■690 ▼a0219
■690 ▼a0485
■690 ▼a0786
■690 ▼a0431
■71020▼aNew York University▼bChemistry.
■7730 ▼tDissertations Abstracts International▼g86-12B.
■790 ▼a0146
■791 ▼aPh.D.
■792 ▼a2025
■793 ▼aEnglish
■85640▼uhttp://www.riss.kr/pdu/ddodLink.do?id=T17356848▼nKERIS▼z이 자료의 원문은 한국교육학술정보원에서 제공합니다.
