Package org.biojava.nbio.core.sequence.io

Class GenbankSequenceParser<S extends AbstractSequence<C>,C extends Compound>

java.lang.Object

org.biojava.nbio.core.sequence.io.GenbankSequenceParser<S,C>

All Implemented Interfaces:

SequenceParserInterface

public class GenbankSequenceParser<S extends AbstractSequence<C>,C extends Compound>
extends Object
implements SequenceParserInterface

Field Summary

Fields

Modifier and Type	Field	Description
protected static String	ACCESSION_TAG
protected static String	AUTHORS_TAG
protected static String	BASE_COUNT_TAG
protected static String	BASE_COUNT_TAG_FULL
protected static String	COMMENT_TAG
protected static String	CONSORTIUM_TAG
protected static String	DBLINK
protected static String	DBSOURCE
protected static Pattern	dbxp
protected static String	DEFINITION_TAG
protected static String	END_SEQUENCE_TAG
protected static String	FEATURE_TAG
static String	GENBANK_FORMAT	The name of this format
protected static Pattern	headerLine
protected static String	JOURNAL_TAG
protected static String	KEYWORDS_TAG
protected static InsdcParser	locationParser
protected static String	LOCUS_TAG
protected static Pattern	lp
protected static Pattern	lp2
protected static String	MEDLINE_TAG
protected static String	ORGANISM_TAG
protected static String	PRIMARY
protected static String	PUBMED_TAG
protected static Pattern	readableFiles
protected static String	REFERENCE_TAG
protected static Pattern	refp
protected static Pattern	refRange
protected static String	REMARK_TAG
protected static Pattern	sectp	sections start at a line and continue till the first line afterwards with a non-whitespace first character we want to match any of the following as a new section within a section \s{0,8} word \s{0,7} value \s{21} /word = value \s{21} /word
protected static String	SOURCE_TAG
protected static String	START_SEQUENCE_TAG
protected static String	TITLE_TAG
protected static String	VERSION_TAG
protected static Pattern	vp

Constructor Summary

Constructors

Constructor	Description
GenbankSequenceParser()

Method Summary

Modifier and Type	Method	Description
CompoundSet<?>	getCompoundType()
Map<String,List<DBReferenceInfo>>	getDatabaseReferences()
Map<String,List<AbstractFeature<AbstractSequence<C>,C>>>	getFeatures()
List<AbstractFeature<AbstractSequence<C>,C>>	getFeatures(String keyword)
String	getHeader()
List<String>	getKeyWords()
String	getSequence(BufferedReader bufferedReader, int sequenceLength)
GenericGenbankHeaderParser<S,C>	getSequenceHeaderParser()
void	parseFeatures(AbstractSequence<C> sequence)

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Field Detail

GENBANK_FORMAT

public static final String GENBANK_FORMAT

The name of this format

See Also:

Constant Field Values

LOCUS_TAG

protected static final String LOCUS_TAG

See Also:

Constant Field Values

DEFINITION_TAG

protected static final String DEFINITION_TAG

See Also:

Constant Field Values

ACCESSION_TAG

protected static final String ACCESSION_TAG

See Also:

Constant Field Values

VERSION_TAG

protected static final String VERSION_TAG

See Also:

Constant Field Values

KEYWORDS_TAG

protected static final String KEYWORDS_TAG

See Also:

Constant Field Values

SOURCE_TAG

protected static final String SOURCE_TAG

See Also:

Constant Field Values

ORGANISM_TAG

protected static final String ORGANISM_TAG

See Also:

Constant Field Values

REFERENCE_TAG

protected static final String REFERENCE_TAG

See Also:

Constant Field Values

AUTHORS_TAG

protected static final String AUTHORS_TAG

See Also:

Constant Field Values

CONSORTIUM_TAG

protected static final String CONSORTIUM_TAG

See Also:

Constant Field Values

TITLE_TAG

protected static final String TITLE_TAG

See Also:

Constant Field Values

JOURNAL_TAG

protected static final String JOURNAL_TAG

See Also:

Constant Field Values

PUBMED_TAG

protected static final String PUBMED_TAG

See Also:

Constant Field Values

MEDLINE_TAG

protected static final String MEDLINE_TAG

See Also:

Constant Field Values

REMARK_TAG

protected static final String REMARK_TAG

See Also:

Constant Field Values

COMMENT_TAG

protected static final String COMMENT_TAG

See Also:

Constant Field Values

FEATURE_TAG

protected static final String FEATURE_TAG

See Also:

Constant Field Values

BASE_COUNT_TAG_FULL

protected static final String BASE_COUNT_TAG_FULL

See Also:

Constant Field Values

BASE_COUNT_TAG

protected static final String BASE_COUNT_TAG

See Also:

Constant Field Values

START_SEQUENCE_TAG

protected static final String START_SEQUENCE_TAG

See Also:

Constant Field Values

DBSOURCE

protected static final String DBSOURCE

See Also:

Constant Field Values

PRIMARY

protected static final String PRIMARY

See Also:

Constant Field Values

DBLINK

protected static final String DBLINK

See Also:

Constant Field Values

END_SEQUENCE_TAG

protected static final String END_SEQUENCE_TAG

See Also:

Constant Field Values

lp

protected static final Pattern lp

lp2

protected static final Pattern lp2

vp

protected static final Pattern vp

refRange

protected static final Pattern refRange

refp

protected static final Pattern refp

dbxp

protected static final Pattern dbxp

locationParser

protected static final InsdcParser locationParser

sectp

protected static final Pattern sectp

sections start at a line and continue till the first line afterwards with a non-whitespace first character we want to match any of the following as a new section within a section \s{0,8} word \s{0,7} value \s{21} /word = value \s{21} /word

readableFiles

protected static final Pattern readableFiles

headerLine

protected static final Pattern headerLine

Constructor Detail

GenbankSequenceParser

public GenbankSequenceParser()

Method Detail

getSequence

public String getSequence(BufferedReader bufferedReader,
                          int sequenceLength)

Specified by:

getSequence in interface SequenceParserInterface

Returns:

getHeader

public String getHeader()

getSequenceHeaderParser

public GenericGenbankHeaderParser<S,C> getSequenceHeaderParser()

getDatabaseReferences

public Map<String,List<DBReferenceInfo>> getDatabaseReferences()

getKeyWords

public List<String> getKeyWords()

getFeatures

public List<AbstractFeature<AbstractSequence<C>,C>> getFeatures(String keyword)

getFeatures

public Map<String,List<AbstractFeature<AbstractSequence<C>,C>>> getFeatures()

parseFeatures

public void parseFeatures(AbstractSequence<C> sequence)

getCompoundType

public CompoundSet<?> getCompoundType()