Profile-based string kernels for remote homology detection and motif extraction.
Identifieur interne : 002316 ( PubMed/Corpus ); précédent : 002315; suivant : 002317Profile-based string kernels for remote homology detection and motif extraction.
Auteurs : Rui Kuang ; Eugene Ie ; Ke Wang ; Kai Wang ; Mahira Siddiqi ; Yoav Freund ; Christina LeslieSource :
- Journal of bioinformatics and computational biology [ 0219-7200 ] ; 2005.
English descriptors
- KwdEn :
- Algorithms, Amino Acid Motifs, Amino Acid Sequence, Artificial Intelligence, Molecular Sequence Data, Pattern Recognition, Automated (methods), Proteins (analysis), Proteins (chemistry), Proteins (classification), Sequence Alignment (methods), Sequence Analysis, Protein (methods), Sequence Homology, Amino Acid.
- MESH :
- chemical , analysis : Proteins.
- chemical , chemistry : Proteins.
- chemical , classification : Proteins.
- methods : Pattern Recognition, Automated, Sequence Alignment, Sequence Analysis, Protein.
- Algorithms, Amino Acid Motifs, Amino Acid Sequence, Artificial Intelligence, Molecular Sequence Data, Sequence Homology, Amino Acid.
Abstract
We introduce novel profile-based string kernels for use with support vector machines (SVMs) for the problems of protein classification and remote homology detection. These kernels use probabilistic profiles, such as those produced by the PSI-BLAST algorithm, to define position-dependent mutation neighborhoods along protein sequences for inexact matching of k-length subsequences ("k-mers") in the data. By use of an efficient data structure, the kernels are fast to compute once the profiles have been obtained. For example, the time needed to run PSI-BLAST in order to build the profiles is significantly longer than both the kernel computation time and the SVM training time. We present remote homology detection experiments based on the SCOP database where we show that profile-based string kernels used with SVM classifiers strongly outperform all recently presented supervised SVM methods. We further examine how to incorporate predicted secondary structure information into the profile kernel to obtain a small but significant performance improvement. We also show how we can use the learned SVM classifier to extract "discriminative sequence motifs"--short regions of the original profile that contribute almost all the weight of the SVM classification score--and show that these discriminative motifs correspond to meaningful structural features in the protein data. The use of PSI-BLAST profiles can be seen as a semi-supervised learning technique, since PSI-BLAST leverages unlabeled data from a large sequence database to build more informative profiles. Recently presented "cluster kernels" give general semi-supervised methods for improving SVM protein classification performance. We show that our profile kernel results also outperform cluster kernels while providing much better scalability to large datasets.
DOI: 10.1142/s021972000500120x
PubMed: 16108083
Links to Exploration step
pubmed:16108083Le document en format XML
<record><TEI><teiHeader><fileDesc><titleStmt><title xml:lang="en">Profile-based string kernels for remote homology detection and motif extraction.</title>
<author><name sortKey="Kuang, Rui" sort="Kuang, Rui" uniqKey="Kuang R" first="Rui" last="Kuang">Rui Kuang</name>
<affiliation><nlm:affiliation>Department of Computer Science, Columbia University, 475 Riverside Dr., Mail Code 7717, New York, NY 10115, USA.</nlm:affiliation>
</affiliation>
</author>
<author><name sortKey="Ie, Eugene" sort="Ie, Eugene" uniqKey="Ie E" first="Eugene" last="Ie">Eugene Ie</name>
</author>
<author><name sortKey="Wang, Ke" sort="Wang, Ke" uniqKey="Wang K" first="Ke" last="Wang">Ke Wang</name>
</author>
<author><name sortKey="Wang, Kai" sort="Wang, Kai" uniqKey="Wang K" first="Kai" last="Wang">Kai Wang</name>
</author>
<author><name sortKey="Siddiqi, Mahira" sort="Siddiqi, Mahira" uniqKey="Siddiqi M" first="Mahira" last="Siddiqi">Mahira Siddiqi</name>
</author>
<author><name sortKey="Freund, Yoav" sort="Freund, Yoav" uniqKey="Freund Y" first="Yoav" last="Freund">Yoav Freund</name>
</author>
<author><name sortKey="Leslie, Christina" sort="Leslie, Christina" uniqKey="Leslie C" first="Christina" last="Leslie">Christina Leslie</name>
</author>
</titleStmt>
<publicationStmt><idno type="wicri:source">PubMed</idno>
<date when="2005">2005</date>
<idno type="RBID">pubmed:16108083</idno>
<idno type="pmid">16108083</idno>
<idno type="doi">10.1142/s021972000500120x</idno>
<idno type="wicri:Area/PubMed/Corpus">002316</idno>
<idno type="wicri:explorRef" wicri:stream="PubMed" wicri:step="Corpus" wicri:corpus="PubMed">002316</idno>
</publicationStmt>
<sourceDesc><biblStruct><analytic><title xml:lang="en">Profile-based string kernels for remote homology detection and motif extraction.</title>
<author><name sortKey="Kuang, Rui" sort="Kuang, Rui" uniqKey="Kuang R" first="Rui" last="Kuang">Rui Kuang</name>
<affiliation><nlm:affiliation>Department of Computer Science, Columbia University, 475 Riverside Dr., Mail Code 7717, New York, NY 10115, USA.</nlm:affiliation>
</affiliation>
</author>
<author><name sortKey="Ie, Eugene" sort="Ie, Eugene" uniqKey="Ie E" first="Eugene" last="Ie">Eugene Ie</name>
</author>
<author><name sortKey="Wang, Ke" sort="Wang, Ke" uniqKey="Wang K" first="Ke" last="Wang">Ke Wang</name>
</author>
<author><name sortKey="Wang, Kai" sort="Wang, Kai" uniqKey="Wang K" first="Kai" last="Wang">Kai Wang</name>
</author>
<author><name sortKey="Siddiqi, Mahira" sort="Siddiqi, Mahira" uniqKey="Siddiqi M" first="Mahira" last="Siddiqi">Mahira Siddiqi</name>
</author>
<author><name sortKey="Freund, Yoav" sort="Freund, Yoav" uniqKey="Freund Y" first="Yoav" last="Freund">Yoav Freund</name>
</author>
<author><name sortKey="Leslie, Christina" sort="Leslie, Christina" uniqKey="Leslie C" first="Christina" last="Leslie">Christina Leslie</name>
</author>
</analytic>
<series><title level="j">Journal of bioinformatics and computational biology</title>
<idno type="ISSN">0219-7200</idno>
<imprint><date when="2005" type="published">2005</date>
</imprint>
</series>
</biblStruct>
</sourceDesc>
</fileDesc>
<profileDesc><textClass><keywords scheme="KwdEn" xml:lang="en"><term>Algorithms</term>
<term>Amino Acid Motifs</term>
<term>Amino Acid Sequence</term>
<term>Artificial Intelligence</term>
<term>Molecular Sequence Data</term>
<term>Pattern Recognition, Automated (methods)</term>
<term>Proteins (analysis)</term>
<term>Proteins (chemistry)</term>
<term>Proteins (classification)</term>
<term>Sequence Alignment (methods)</term>
<term>Sequence Analysis, Protein (methods)</term>
<term>Sequence Homology, Amino Acid</term>
</keywords>
<keywords scheme="MESH" type="chemical" qualifier="analysis" xml:lang="en"><term>Proteins</term>
</keywords>
<keywords scheme="MESH" type="chemical" qualifier="chemistry" xml:lang="en"><term>Proteins</term>
</keywords>
<keywords scheme="MESH" type="chemical" qualifier="classification" xml:lang="en"><term>Proteins</term>
</keywords>
<keywords scheme="MESH" qualifier="methods" xml:lang="en"><term>Pattern Recognition, Automated</term>
<term>Sequence Alignment</term>
<term>Sequence Analysis, Protein</term>
</keywords>
<keywords scheme="MESH" xml:lang="en"><term>Algorithms</term>
<term>Amino Acid Motifs</term>
<term>Amino Acid Sequence</term>
<term>Artificial Intelligence</term>
<term>Molecular Sequence Data</term>
<term>Sequence Homology, Amino Acid</term>
</keywords>
</textClass>
</profileDesc>
</teiHeader>
<front><div type="abstract" xml:lang="en">We introduce novel profile-based string kernels for use with support vector machines (SVMs) for the problems of protein classification and remote homology detection. These kernels use probabilistic profiles, such as those produced by the PSI-BLAST algorithm, to define position-dependent mutation neighborhoods along protein sequences for inexact matching of k-length subsequences ("k-mers") in the data. By use of an efficient data structure, the kernels are fast to compute once the profiles have been obtained. For example, the time needed to run PSI-BLAST in order to build the profiles is significantly longer than both the kernel computation time and the SVM training time. We present remote homology detection experiments based on the SCOP database where we show that profile-based string kernels used with SVM classifiers strongly outperform all recently presented supervised SVM methods. We further examine how to incorporate predicted secondary structure information into the profile kernel to obtain a small but significant performance improvement. We also show how we can use the learned SVM classifier to extract "discriminative sequence motifs"--short regions of the original profile that contribute almost all the weight of the SVM classification score--and show that these discriminative motifs correspond to meaningful structural features in the protein data. The use of PSI-BLAST profiles can be seen as a semi-supervised learning technique, since PSI-BLAST leverages unlabeled data from a large sequence database to build more informative profiles. Recently presented "cluster kernels" give general semi-supervised methods for improving SVM protein classification performance. We show that our profile kernel results also outperform cluster kernels while providing much better scalability to large datasets.</div>
</front>
</TEI>
<pubmed><MedlineCitation Status="MEDLINE" Owner="NLM"><PMID Version="1">16108083</PMID>
<DateCompleted><Year>2005</Year>
<Month>09</Month>
<Day>23</Day>
</DateCompleted>
<DateRevised><Year>2019</Year>
<Month>12</Month>
<Day>10</Day>
</DateRevised>
<Article PubModel="Print"><Journal><ISSN IssnType="Print">0219-7200</ISSN>
<JournalIssue CitedMedium="Print"><Volume>3</Volume>
<Issue>3</Issue>
<PubDate><Year>2005</Year>
<Month>Jun</Month>
</PubDate>
</JournalIssue>
<Title>Journal of bioinformatics and computational biology</Title>
<ISOAbbreviation>J Bioinform Comput Biol</ISOAbbreviation>
</Journal>
<ArticleTitle>Profile-based string kernels for remote homology detection and motif extraction.</ArticleTitle>
<Pagination><MedlinePgn>527-50</MedlinePgn>
</Pagination>
<Abstract><AbstractText>We introduce novel profile-based string kernels for use with support vector machines (SVMs) for the problems of protein classification and remote homology detection. These kernels use probabilistic profiles, such as those produced by the PSI-BLAST algorithm, to define position-dependent mutation neighborhoods along protein sequences for inexact matching of k-length subsequences ("k-mers") in the data. By use of an efficient data structure, the kernels are fast to compute once the profiles have been obtained. For example, the time needed to run PSI-BLAST in order to build the profiles is significantly longer than both the kernel computation time and the SVM training time. We present remote homology detection experiments based on the SCOP database where we show that profile-based string kernels used with SVM classifiers strongly outperform all recently presented supervised SVM methods. We further examine how to incorporate predicted secondary structure information into the profile kernel to obtain a small but significant performance improvement. We also show how we can use the learned SVM classifier to extract "discriminative sequence motifs"--short regions of the original profile that contribute almost all the weight of the SVM classification score--and show that these discriminative motifs correspond to meaningful structural features in the protein data. The use of PSI-BLAST profiles can be seen as a semi-supervised learning technique, since PSI-BLAST leverages unlabeled data from a large sequence database to build more informative profiles. Recently presented "cluster kernels" give general semi-supervised methods for improving SVM protein classification performance. We show that our profile kernel results also outperform cluster kernels while providing much better scalability to large datasets.</AbstractText>
</Abstract>
<AuthorList CompleteYN="Y"><Author ValidYN="Y"><LastName>Kuang</LastName>
<ForeName>Rui</ForeName>
<Initials>R</Initials>
<AffiliationInfo><Affiliation>Department of Computer Science, Columbia University, 475 Riverside Dr., Mail Code 7717, New York, NY 10115, USA.</Affiliation>
</AffiliationInfo>
</Author>
<Author ValidYN="Y"><LastName>Ie</LastName>
<ForeName>Eugene</ForeName>
<Initials>E</Initials>
</Author>
<Author ValidYN="Y"><LastName>Wang</LastName>
<ForeName>Ke</ForeName>
<Initials>K</Initials>
</Author>
<Author ValidYN="Y"><LastName>Wang</LastName>
<ForeName>Kai</ForeName>
<Initials>K</Initials>
</Author>
<Author ValidYN="Y"><LastName>Siddiqi</LastName>
<ForeName>Mahira</ForeName>
<Initials>M</Initials>
</Author>
<Author ValidYN="Y"><LastName>Freund</LastName>
<ForeName>Yoav</ForeName>
<Initials>Y</Initials>
</Author>
<Author ValidYN="Y"><LastName>Leslie</LastName>
<ForeName>Christina</ForeName>
<Initials>C</Initials>
</Author>
</AuthorList>
<Language>eng</Language>
<GrantList CompleteYN="Y"><Grant><GrantID>LM 07276-02</GrantID>
<Acronym>LM</Acronym>
<Agency>NLM NIH HHS</Agency>
<Country>United States</Country>
</Grant>
</GrantList>
<PublicationTypeList><PublicationType UI="D003160">Comparative Study</PublicationType>
<PublicationType UI="D023362">Evaluation Study</PublicationType>
<PublicationType UI="D016428">Journal Article</PublicationType>
<PublicationType UI="D052061">Research Support, N.I.H., Extramural</PublicationType>
<PublicationType UI="D013485">Research Support, Non-U.S. Gov't</PublicationType>
<PublicationType UI="D013486">Research Support, U.S. Gov't, Non-P.H.S.</PublicationType>
<PublicationType UI="D013487">Research Support, U.S. Gov't, P.H.S.</PublicationType>
</PublicationTypeList>
</Article>
<MedlineJournalInfo><Country>Singapore</Country>
<MedlineTA>J Bioinform Comput Biol</MedlineTA>
<NlmUniqueID>101187344</NlmUniqueID>
<ISSNLinking>0219-7200</ISSNLinking>
</MedlineJournalInfo>
<ChemicalList><Chemical><RegistryNumber>0</RegistryNumber>
<NameOfSubstance UI="D011506">Proteins</NameOfSubstance>
</Chemical>
</ChemicalList>
<CitationSubset>IM</CitationSubset>
<MeshHeadingList><MeshHeading><DescriptorName UI="D000465" MajorTopicYN="Y">Algorithms</DescriptorName>
</MeshHeading>
<MeshHeading><DescriptorName UI="D020816" MajorTopicYN="N">Amino Acid Motifs</DescriptorName>
</MeshHeading>
<MeshHeading><DescriptorName UI="D000595" MajorTopicYN="N">Amino Acid Sequence</DescriptorName>
</MeshHeading>
<MeshHeading><DescriptorName UI="D001185" MajorTopicYN="Y">Artificial Intelligence</DescriptorName>
</MeshHeading>
<MeshHeading><DescriptorName UI="D008969" MajorTopicYN="N">Molecular Sequence Data</DescriptorName>
</MeshHeading>
<MeshHeading><DescriptorName UI="D010363" MajorTopicYN="N">Pattern Recognition, Automated</DescriptorName>
<QualifierName UI="Q000379" MajorTopicYN="Y">methods</QualifierName>
</MeshHeading>
<MeshHeading><DescriptorName UI="D011506" MajorTopicYN="N">Proteins</DescriptorName>
<QualifierName UI="Q000032" MajorTopicYN="N">analysis</QualifierName>
<QualifierName UI="Q000737" MajorTopicYN="Y">chemistry</QualifierName>
<QualifierName UI="Q000145" MajorTopicYN="N">classification</QualifierName>
</MeshHeading>
<MeshHeading><DescriptorName UI="D016415" MajorTopicYN="N">Sequence Alignment</DescriptorName>
<QualifierName UI="Q000379" MajorTopicYN="Y">methods</QualifierName>
</MeshHeading>
<MeshHeading><DescriptorName UI="D020539" MajorTopicYN="N">Sequence Analysis, Protein</DescriptorName>
<QualifierName UI="Q000379" MajorTopicYN="Y">methods</QualifierName>
</MeshHeading>
<MeshHeading><DescriptorName UI="D017386" MajorTopicYN="N">Sequence Homology, Amino Acid</DescriptorName>
</MeshHeading>
</MeshHeadingList>
</MedlineCitation>
<PubmedData><History><PubMedPubDate PubStatus="received"><Year>2004</Year>
<Month>09</Month>
<Day>02</Day>
</PubMedPubDate>
<PubMedPubDate PubStatus="revised"><Year>2004</Year>
<Month>12</Month>
<Day>02</Day>
</PubMedPubDate>
<PubMedPubDate PubStatus="accepted"><Year>2004</Year>
<Month>12</Month>
<Day>02</Day>
</PubMedPubDate>
<PubMedPubDate PubStatus="pubmed"><Year>2005</Year>
<Month>8</Month>
<Day>19</Day>
<Hour>9</Hour>
<Minute>0</Minute>
</PubMedPubDate>
<PubMedPubDate PubStatus="medline"><Year>2005</Year>
<Month>9</Month>
<Day>24</Day>
<Hour>9</Hour>
<Minute>0</Minute>
</PubMedPubDate>
<PubMedPubDate PubStatus="entrez"><Year>2005</Year>
<Month>8</Month>
<Day>19</Day>
<Hour>9</Hour>
<Minute>0</Minute>
</PubMedPubDate>
</History>
<PublicationStatus>ppublish</PublicationStatus>
<ArticleIdList><ArticleId IdType="pubmed">16108083</ArticleId>
<ArticleId IdType="pii">S021972000500120X</ArticleId>
<ArticleId IdType="doi">10.1142/s021972000500120x</ArticleId>
</ArticleIdList>
</PubmedData>
</pubmed>
</record>
Pour manipuler ce document sous Unix (Dilib)
EXPLOR_STEP=$WICRI_ROOT/Sante/explor/MersV1/Data/PubMed/Corpus
HfdSelect -h $EXPLOR_STEP/biblio.hfd -nk 002316 | SxmlIndent | more
Ou
HfdSelect -h $EXPLOR_AREA/Data/PubMed/Corpus/biblio.hfd -nk 002316 | SxmlIndent | more
Pour mettre un lien sur cette page dans le réseau Wicri
{{Explor lien |wiki= Sante |area= MersV1 |flux= PubMed |étape= Corpus |type= RBID |clé= pubmed:16108083 |texte= Profile-based string kernels for remote homology detection and motif extraction. }}
Pour générer des pages wiki
HfdIndexSelect -h $EXPLOR_AREA/Data/PubMed/Corpus/RBID.i -Sk "pubmed:16108083" \ | HfdSelect -Kh $EXPLOR_AREA/Data/PubMed/Corpus/biblio.hfd \ | NlmPubMed2Wicri -a MersV1
![]() | This area was generated with Dilib version V0.6.33. | ![]() |