Simplified sequence-based method for ATP-binding prediction using contextual local evolutionary conservation

Chun Fang, Tamotsu Noguchi, Hayato Yamana

    Research output: Contribution to journalArticle

    6 Citations (Scopus)

    Abstract

    Background: Identifying ligand-binding sites is a key step to annotate the protein functions and to find applications in drug design. Now, many sequence-based methods adopted various predicted results from other classifiers, such as predicted secondary structure, predicted solvent accessibility and predicted disorder probabilities, to combine with position-specific scoring matrix (PSSM) as input for binding sites prediction. These predicted features not only easily result in high-dimensional feature space, but also greatly increased the complexity of algorithms. Moreover, the performances of these predictors are also largely influenced by the other classifiers.Results: In order to verify that conservation is the most powerful attribute in identifying ligand-binding sites, and to show the importance of revising PSSM to match the detailed conservation pattern of functional site in prediction, we have analyzed the Adenosine-5'-triphosphate (ATP) ligand as an example, and proposed a simple method for ATP-binding sites prediction, named as CLCLpred (Contextual Local evolutionary Conservation-based method for Ligand-binding prediction). Our method employed no predicted results from other classifiers as input; all used features were extracted from PSSM only. We tested our method on 2 separate data sets. Experimental results showed that, comparing with other 9 existing methods on the same data sets, our method achieved the best performance.Conclusions: This study demonstrates that: 1) exploiting the signal from the detailed conservation pattern of residues will largely facilitate the prediction of protein functional sites; and 2) the local evolutionary conservation enables accurate prediction of ATP-binding sites directly from protein sequence.

    Original languageEnglish
    Article number7
    JournalAlgorithms for Molecular Biology
    Volume9
    Issue number1
    DOIs
    Publication statusPublished - 2014 Mar 11

    Fingerprint

    Adenosine
    Binding sites
    Conservation
    Adenosine Triphosphate
    Position-Specific Scoring Matrices
    Prediction
    Binding Sites
    Ligands
    Scoring
    Classifiers
    Proteins
    Classifier
    Protein
    Drug Design
    Protein Sequence
    Secondary Structure
    Feature Space
    Accessibility
    Disorder
    Predictors

    Keywords

    • ATP-binding site
    • Local evolutionary conservation
    • Sequence-based

    ASJC Scopus subject areas

    • Computational Theory and Mathematics
    • Applied Mathematics
    • Molecular Biology
    • Structural Biology

    Cite this

    Simplified sequence-based method for ATP-binding prediction using contextual local evolutionary conservation. / Fang, Chun; Noguchi, Tamotsu; Yamana, Hayato.

    In: Algorithms for Molecular Biology, Vol. 9, No. 1, 7, 11.03.2014.

    Research output: Contribution to journalArticle

    @article{9828f06aae2948e485d9a9d48b3d5661,
    title = "Simplified sequence-based method for ATP-binding prediction using contextual local evolutionary conservation",
    abstract = "Background: Identifying ligand-binding sites is a key step to annotate the protein functions and to find applications in drug design. Now, many sequence-based methods adopted various predicted results from other classifiers, such as predicted secondary structure, predicted solvent accessibility and predicted disorder probabilities, to combine with position-specific scoring matrix (PSSM) as input for binding sites prediction. These predicted features not only easily result in high-dimensional feature space, but also greatly increased the complexity of algorithms. Moreover, the performances of these predictors are also largely influenced by the other classifiers.Results: In order to verify that conservation is the most powerful attribute in identifying ligand-binding sites, and to show the importance of revising PSSM to match the detailed conservation pattern of functional site in prediction, we have analyzed the Adenosine-5'-triphosphate (ATP) ligand as an example, and proposed a simple method for ATP-binding sites prediction, named as CLCLpred (Contextual Local evolutionary Conservation-based method for Ligand-binding prediction). Our method employed no predicted results from other classifiers as input; all used features were extracted from PSSM only. We tested our method on 2 separate data sets. Experimental results showed that, comparing with other 9 existing methods on the same data sets, our method achieved the best performance.Conclusions: This study demonstrates that: 1) exploiting the signal from the detailed conservation pattern of residues will largely facilitate the prediction of protein functional sites; and 2) the local evolutionary conservation enables accurate prediction of ATP-binding sites directly from protein sequence.",
    keywords = "ATP-binding site, Local evolutionary conservation, Sequence-based",
    author = "Chun Fang and Tamotsu Noguchi and Hayato Yamana",
    year = "2014",
    month = "3",
    day = "11",
    doi = "10.1186/1748-7188-9-7",
    language = "English",
    volume = "9",
    journal = "Algorithms for Molecular Biology",
    issn = "1748-7188",
    publisher = "BioMed Central",
    number = "1",

    }

    TY - JOUR

    T1 - Simplified sequence-based method for ATP-binding prediction using contextual local evolutionary conservation

    AU - Fang, Chun

    AU - Noguchi, Tamotsu

    AU - Yamana, Hayato

    PY - 2014/3/11

    Y1 - 2014/3/11

    N2 - Background: Identifying ligand-binding sites is a key step to annotate the protein functions and to find applications in drug design. Now, many sequence-based methods adopted various predicted results from other classifiers, such as predicted secondary structure, predicted solvent accessibility and predicted disorder probabilities, to combine with position-specific scoring matrix (PSSM) as input for binding sites prediction. These predicted features not only easily result in high-dimensional feature space, but also greatly increased the complexity of algorithms. Moreover, the performances of these predictors are also largely influenced by the other classifiers.Results: In order to verify that conservation is the most powerful attribute in identifying ligand-binding sites, and to show the importance of revising PSSM to match the detailed conservation pattern of functional site in prediction, we have analyzed the Adenosine-5'-triphosphate (ATP) ligand as an example, and proposed a simple method for ATP-binding sites prediction, named as CLCLpred (Contextual Local evolutionary Conservation-based method for Ligand-binding prediction). Our method employed no predicted results from other classifiers as input; all used features were extracted from PSSM only. We tested our method on 2 separate data sets. Experimental results showed that, comparing with other 9 existing methods on the same data sets, our method achieved the best performance.Conclusions: This study demonstrates that: 1) exploiting the signal from the detailed conservation pattern of residues will largely facilitate the prediction of protein functional sites; and 2) the local evolutionary conservation enables accurate prediction of ATP-binding sites directly from protein sequence.

    AB - Background: Identifying ligand-binding sites is a key step to annotate the protein functions and to find applications in drug design. Now, many sequence-based methods adopted various predicted results from other classifiers, such as predicted secondary structure, predicted solvent accessibility and predicted disorder probabilities, to combine with position-specific scoring matrix (PSSM) as input for binding sites prediction. These predicted features not only easily result in high-dimensional feature space, but also greatly increased the complexity of algorithms. Moreover, the performances of these predictors are also largely influenced by the other classifiers.Results: In order to verify that conservation is the most powerful attribute in identifying ligand-binding sites, and to show the importance of revising PSSM to match the detailed conservation pattern of functional site in prediction, we have analyzed the Adenosine-5'-triphosphate (ATP) ligand as an example, and proposed a simple method for ATP-binding sites prediction, named as CLCLpred (Contextual Local evolutionary Conservation-based method for Ligand-binding prediction). Our method employed no predicted results from other classifiers as input; all used features were extracted from PSSM only. We tested our method on 2 separate data sets. Experimental results showed that, comparing with other 9 existing methods on the same data sets, our method achieved the best performance.Conclusions: This study demonstrates that: 1) exploiting the signal from the detailed conservation pattern of residues will largely facilitate the prediction of protein functional sites; and 2) the local evolutionary conservation enables accurate prediction of ATP-binding sites directly from protein sequence.

    KW - ATP-binding site

    KW - Local evolutionary conservation

    KW - Sequence-based

    UR - http://www.scopus.com/inward/record.url?scp=84899131695&partnerID=8YFLogxK

    UR - http://www.scopus.com/inward/citedby.url?scp=84899131695&partnerID=8YFLogxK

    U2 - 10.1186/1748-7188-9-7

    DO - 10.1186/1748-7188-9-7

    M3 - Article

    AN - SCOPUS:84899131695

    VL - 9

    JO - Algorithms for Molecular Biology

    JF - Algorithms for Molecular Biology

    SN - 1748-7188

    IS - 1

    M1 - 7

    ER -