YEARS

2006-2012

AUTHORS

Jude W Shavlik

TITLE

Machine Learning and Visualization in Structural Biology

ABSTRACT

DESCRIPTION: This project's main objective is to develop computerized tools that assist x-ray crystallographers in rapidly determining the three-dimensional structure of a protein. More specifically, this project addresses the following task: given a 3D electron-density map from crystallography and the sequence of the protein, find the most likely layout (i.e. "trace") of the protein sequence in 3D. The project will create both automated methods based on statistical machine-learning and computer-vision techniques, as well as visualization tools that support humans doing this layout. These two approaches complement each other and are synergistic. This project's first specific aim is to develop and empirically evaluate algorithms that interpret crystallographic electron-density maps. The second specific aim is to incorporate structural-biology domain knowledge (secondary-structure prediction and potential-energy calculations) into the project's algorithms for interpreting density maps. The third specific aim is to tightly integrate partial model-construction with phase estimation updates to improve the recognition of 3D protein structures in x-ray reflection data;crystallographers will be able to intervene whenever they desire to help "steer" this iterative process. The final specific aim is to develop intuitive and effective modalities - including virtual reality and the use of speech/audio - for the efficient use of crystallographer's time in manual model fitting and validation. Structural biology has wide relevance to biomedicine, since protein function generally follows from protein form (i.e., its structure). This project's techniques will speed-up the process of determining protein 3D structures, especially from low-quality (i.e., low-resolution) x-ray data, and will be applicable to other structural-biology tasks. Being able to accurately interpret low-resolution data promises to allow higher through put structure determination. The broader impact will include a better understanding of the power of modern theories and algorithms in machine learning and visualization in solving biological problems.

FUNDED PUBLICATIONS

  • Structural characterization of human Uch37.
  • Creating protein models from electron-density maps using particle-filtering methods.
  • A probabilistic approach to protein backbone tracing in electron density maps.
  • Spherical-harmonic decomposition for molecular recognition in electron-density maps.
  • Probabilistic ensembles for improved inference in protein-structure determination.
  • How to use: Click on a object to move its position. Double click to open its homepage. Right click to preview its contents.

    Download the RDF metadata as:   json-ld nt turtle xml License info


    22 TRIPLES      17 PREDICATES      23 URIs      9 LITERALS

    Subject Predicate Object
    1 grants:52d91ba2ee62dd696d65cbe8a25ca952 sg:abstract DESCRIPTION: This project's main objective is to develop computerized tools that assist x-ray crystallographers in rapidly determining the three-dimensional structure of a protein. More specifically, this project addresses the following task: given a 3D electron-density map from crystallography and the sequence of the protein, find the most likely layout (i.e. "trace") of the protein sequence in 3D. The project will create both automated methods based on statistical machine-learning and computer-vision techniques, as well as visualization tools that support humans doing this layout. These two approaches complement each other and are synergistic. This project's first specific aim is to develop and empirically evaluate algorithms that interpret crystallographic electron-density maps. The second specific aim is to incorporate structural-biology domain knowledge (secondary-structure prediction and potential-energy calculations) into the project's algorithms for interpreting density maps. The third specific aim is to tightly integrate partial model-construction with phase estimation updates to improve the recognition of 3D protein structures in x-ray reflection data;crystallographers will be able to intervene whenever they desire to help "steer" this iterative process. The final specific aim is to develop intuitive and effective modalities - including virtual reality and the use of speech/audio - for the efficient use of crystallographer's time in manual model fitting and validation. Structural biology has wide relevance to biomedicine, since protein function generally follows from protein form (i.e., its structure). This project's techniques will speed-up the process of determining protein 3D structures, especially from low-quality (i.e., low-resolution) x-ray data, and will be applicable to other structural-biology tasks. Being able to accurately interpret low-resolution data promises to allow higher through put structure determination. The broader impact will include a better understanding of the power of modern theories and algorithms in machine learning and visualization in solving biological problems.
    2 sg:endYear 2012
    3 sg:fundingAmount 1249603.0
    4 sg:fundingCurrency USD
    5 sg:hasContribution contributions:32ae843c1291f477cec5adee270f76eb
    6 sg:hasFieldOfResearchCode anzsrc-for:08
    7 anzsrc-for:0801
    8 sg:hasFundedPublication articles:38974a8c373ada2c7238ec7d389d6276
    9 articles:52e300aff7989157b71b3028e641eb5f
    10 articles:873567000f50eb707c8c4530dd00c107
    11 articles:bbf6329a09d57cd3f694fa500d1cc7fc
    12 articles:e1b90236823225b516751cc7c960b93d
    13 sg:hasFundingOrganization grid-institutes:grid.280285.5
    14 sg:hasRecipientOrganization grid-institutes:grid.14003.36
    15 sg:language English
    16 sg:license http://scigraph.springernature.com/explorer/license/
    17 sg:scigraphId 52d91ba2ee62dd696d65cbe8a25ca952
    18 sg:startYear 2006
    19 sg:title Machine Learning and Visualization in Structural Biology
    20 sg:webpage http://projectreporter.nih.gov/project_info_description.cfm?aid=7599114
    21 rdf:type sg:Grant
    22 rdfs:label Grant: Machine Learning and Visualization in Structural Biology
    HOW TO GET THIS DATA PROGRAMMATICALLY:

    JSON-LD is a popular JSON format for linked data.

    curl -H 'Accept: application/ld+json' 'http://scigraph.springernature.com/things/grants/52d91ba2ee62dd696d65cbe8a25ca952'

    N-Triples is a line-based linked data format ideal for batch operations .

    curl -H 'Accept: application/n-triples' 'http://scigraph.springernature.com/things/grants/52d91ba2ee62dd696d65cbe8a25ca952'

    Turtle is a human-readable linked data format.

    curl -H 'Accept: text/turtle' 'http://scigraph.springernature.com/things/grants/52d91ba2ee62dd696d65cbe8a25ca952'

    RDF/XML is a standard XML format for linked data.

    curl -H 'Accept: application/rdf+xml' 'http://scigraph.springernature.com/things/grants/52d91ba2ee62dd696d65cbe8a25ca952'






    Preview window. Press ESC to close (or click here)


    ...