import org.forester.evoinference.distance.PairwiseDistanceCalculator.PWD_DISTANCE_METHOD;
import org.forester.evoinference.matrix.distance.BasicSymmetricalDistanceMatrix;
import org.forester.evoinference.tools.BootstrapResampler;
-import org.forester.msa.BasicMsa;
+import org.forester.io.writers.SequenceWriter;
+import org.forester.io.writers.SequenceWriter.SEQ_FORMAT;
import org.forester.msa.DeleteableMsa;
import org.forester.msa.Mafft;
import org.forester.msa.Msa;
final private static NumberFormat NF_3 = new DecimalFormat( "#.###" );
final private static NumberFormat NF_4 = new DecimalFormat( "#.####" );
private double _gap_ratio = -1;
+ private final short _longest_id_length;
//
- private final String _maffts_opts = "--auto";
+ private String _maffts_opts = "--auto";
private int _min_length = -1;
//
private DeleteableMsa _msa = null;
//
private boolean _realign = false;
private final SortedSet<String> _removed_seq_ids;
- private final File _removed_seqs_out_base = null;
+ private final ArrayList<Sequence> _removed_seqs;
+ private File _removed_seqs_out_base = null;
private boolean _report_aln_mean_identity = false;
private int _step = -1;
private int _step_for_diagnostics = -1;
- private final short _longest_id_length;
- private final ArrayList<Sequence> _removed_seqs;
static {
NF_4.setRoundingMode( RoundingMode.HALF_UP );
NF_3.setRoundingMode( RoundingMode.HALF_UP );
_removed_seqs = new ArrayList<Sequence>();
}
+ public final List<MsaProperties> chart( final int step, final boolean realign, final boolean norm )
+ throws IOException, InterruptedException {
+ final GapContribution stats[] = calcGapContribtionsStats( norm );
+ final List<String> to_remove_ids = new ArrayList<String>();
+ final List<MsaProperties> msa_props = new ArrayList<MsaProperties>();
+ for( final GapContribution gap_gontribution : stats ) {
+ to_remove_ids.add( gap_gontribution.getId() );
+ }
+ if ( !_realign ) {
+ _step = -1;
+ }
+ printTableHeader();
+ int x = ForesterUtil.roundToInt( _msa.getNumberOfSequences() / 20.0 );
+ if ( x < 1 ) {
+ x = 1;
+ }
+ MsaProperties msa_prop = new MsaProperties( _msa, _report_aln_mean_identity );
+ msa_props.add( msa_prop );
+ printMsaProperties( "", msa_prop );
+ System.out.println();
+ int i = 0;
+ while ( _msa.getNumberOfSequences() > x ) {
+ final String id = to_remove_ids.get( i );
+ _msa.deleteRow( id, false );
+ if ( realign && isPrintMsaStatsWriteOutfileAndRealign( i ) ) {
+ removeGapColumns();
+ realignWithMafft();
+ msa_prop = new MsaProperties( _msa, _report_aln_mean_identity );
+ msa_props.add( msa_prop );
+ printMsaProperties( id, msa_prop );
+ System.out.print( "(realigned)" );
+ System.out.println();
+ }
+ else if ( isPrintMsaStats( i ) ) {
+ removeGapColumns();
+ msa_prop = new MsaProperties( _msa, _report_aln_mean_identity );
+ msa_props.add( msa_prop );
+ printMsaProperties( id, msa_prop );
+ System.out.println();
+ }
+ ++i;
+ }
+ return msa_props;
+ }
+
+ final public void deleteGapColumns( final double max_allowed_gap_ratio ) {
+ _msa.deleteGapColumns( max_allowed_gap_ratio );
+ }
+
final public Msa getMsa() {
return _msa;
}
return _removed_seq_ids;
}
- public final void removeViaGapAverage( final double mean_gapiness ) throws IOException, InterruptedException {
+ public final void removeSequencesByMinimalLength( final int min_effective_length ) {
+ printMsaProperties( "", new MsaProperties( _msa, _report_aln_mean_identity ) );
+ System.out.println();
+ _msa = DeleteableMsa.createInstance( MsaMethods.removeSequencesByMinimalLength( _msa, min_effective_length ) );
+ removeGapColumns();
+ printMsaProperties( "", new MsaProperties( _msa, _report_aln_mean_identity ) );
+ System.out.println();
+ }
+
+ public final List<MsaProperties> removeViaGapAverage( final double mean_gapiness ) throws IOException,
+ InterruptedException {
final GapContribution stats[] = calcGapContribtionsStats( _norm );
final List<String> to_remove_ids = new ArrayList<String>();
+ final List<MsaProperties> msa_props = new ArrayList<MsaProperties>();
for( final GapContribution gap_gontribution : stats ) {
to_remove_ids.add( gap_gontribution.getId() );
}
printTableHeader();
+ MsaProperties msa_prop = new MsaProperties( _msa, _report_aln_mean_identity );
+ msa_props.add( msa_prop );
+ printMsaProperties( "", msa_prop );
+ System.out.println();
int i = 0;
while ( MsaMethods.calcGapRatio( _msa ) > mean_gapiness ) {
final String id = to_remove_ids.get( i );
_removed_seq_ids.add( id );
- final Sequence deleted = _msa.deleteRow( id );
+ final Sequence deleted = _msa.deleteRow( id, true );
_removed_seqs.add( deleted );
removeGapColumns();
- if ( ( ( _step > 0 ) && ( ( ( i + 1 ) % _step ) == 0 ) )
- || ( MsaMethods.calcGapRatio( _msa ) <= mean_gapiness ) ) {
- printMsaStatsWriteOutfileAndRealign( _realign, id );
+ if ( isPrintMsaStatsWriteOutfileAndRealign( i ) || ( MsaMethods.calcGapRatio( _msa ) <= mean_gapiness ) ) {
+ msa_prop = printMsaStatsWriteOutfileAndRealign( _realign, id );
+ msa_props.add( msa_prop );
+ System.out.println();
}
- else {
- final MsaProperties msa_prop = new MsaProperties( _msa, _report_aln_mean_identity );
+ else if ( isPrintMsaStats( i ) ) {
+ msa_prop = new MsaProperties( _msa, _report_aln_mean_identity );
+ msa_props.add( msa_prop );
printMsaProperties( id, msa_prop );
+ System.out.println();
}
- System.out.println();
++i;
}
- final String msg = writeAndAlignRemovedSeqs();
- System.out.println( msg );
+ if ( _removed_seqs_out_base != null ) {
+ final String msg = writeAndAlignRemovedSeqs();
+ System.out.println();
+ System.out.println( msg );
+ }
+ return msa_props;
}
- public void removeViaLength( final int length ) throws IOException, InterruptedException {
+ public List<MsaProperties> removeViaLength( final int length ) throws IOException, InterruptedException {
final GapContribution stats[] = calcGapContribtionsStats( _norm );
final List<String> to_remove_ids = new ArrayList<String>();
+ final List<MsaProperties> msa_props = new ArrayList<MsaProperties>();
for( final GapContribution gap_gontribution : stats ) {
to_remove_ids.add( gap_gontribution.getId() );
}
printTableHeader();
+ MsaProperties msa_prop = new MsaProperties( _msa, _report_aln_mean_identity );
+ msa_props.add( msa_prop );
+ printMsaProperties( "", msa_prop );
+ System.out.println();
int i = 0;
while ( _msa.getLength() > length ) {
final String id = to_remove_ids.get( i );
_removed_seq_ids.add( id );
- final Sequence deleted = _msa.deleteRow( id );
- _removed_seqs.add( deleted );
- removeGapColumns();
- if ( ( ( _step > 0 ) && ( ( ( i + 1 ) % _step ) == 0 ) ) || ( _msa.getLength() <= length ) ) {
- printMsaStatsWriteOutfileAndRealign( _realign, id );
- }
- final MsaProperties msa_prop = new MsaProperties( _msa, _report_aln_mean_identity );
- printMsaProperties( id, msa_prop );
- System.out.println();
- ++i;
- }
- final String msg = writeAndAlignRemovedSeqs();
- System.out.println( msg );
- }
-
- public final void removeWorstOffenders( final int to_remove ) throws IOException, InterruptedException {
- final GapContribution stats[] = calcGapContribtionsStats( _norm );
- final List<String> to_remove_ids = new ArrayList<String>();
- for( int j = 0; j < to_remove; ++j ) {
- to_remove_ids.add( stats[ j ].getId() );
- _removed_seq_ids.add( stats[ j ].getId() );
- }
- printTableHeader();
- for( int i = 0; i < to_remove_ids.size(); ++i ) {
- final String id = to_remove_ids.get( i );
- _removed_seq_ids.add( id );
- final Sequence deleted = _msa.deleteRow( id );
+ final Sequence deleted = _msa.deleteRow( id, true );
_removed_seqs.add( deleted );
removeGapColumns();
- if ( isPrintMsaStatsWriteOutfileAndRealign( i ) || ( i == ( to_remove_ids.size() - 1 ) ) ) {
- printMsaStatsWriteOutfileAndRealign( _realign, id );
+ if ( isPrintMsaStatsWriteOutfileAndRealign( i ) || ( _msa.getLength() <= length ) ) {
+ msa_prop = printMsaStatsWriteOutfileAndRealign( _realign, id );
+ msa_props.add( msa_prop );
System.out.println();
}
else if ( isPrintMsaStats( i ) ) {
- final MsaProperties msa_prop = new MsaProperties( _msa, _report_aln_mean_identity );
+ msa_prop = new MsaProperties( _msa, _report_aln_mean_identity );
printMsaProperties( id, msa_prop );
+ msa_props.add( msa_prop );
System.out.println();
}
+ ++i;
}
- final String msg = writeAndAlignRemovedSeqs();
- System.out.println( msg );
+ if ( _removed_seqs_out_base != null ) {
+ final String msg = writeAndAlignRemovedSeqs();
+ System.out.println();
+ System.out.println( msg );
+ }
+ return msa_props;
}
- public final List<MsaProperties> chart( final int step, final boolean realign, final boolean norm )
- throws IOException, InterruptedException {
- final GapContribution stats[] = calcGapContribtionsStats( norm );
+ public final List<MsaProperties> removeWorstOffenders( final int to_remove ) throws IOException,
+ InterruptedException {
+ final GapContribution stats[] = calcGapContribtionsStats( _norm );
final List<String> to_remove_ids = new ArrayList<String>();
final List<MsaProperties> msa_props = new ArrayList<MsaProperties>();
- for( final GapContribution gap_gontribution : stats ) {
- to_remove_ids.add( gap_gontribution.getId() );
+ for( int j = 0; j < to_remove; ++j ) {
+ to_remove_ids.add( stats[ j ].getId() );
+ _removed_seq_ids.add( stats[ j ].getId() );
}
printTableHeader();
- int i = 0;
- final int x = ForesterUtil.roundToInt( _msa.getNumberOfSequences() / 20.0 );
MsaProperties msa_prop = new MsaProperties( _msa, _report_aln_mean_identity );
msa_props.add( msa_prop );
printMsaProperties( "", msa_prop );
System.out.println();
- while ( _msa.getNumberOfSequences() > x ) {
+ for( int i = 0; i < to_remove_ids.size(); ++i ) {
final String id = to_remove_ids.get( i );
- _msa.deleteRow( id );
- if ( realign && isPrintMsaStatsWriteOutfileAndRealign( i ) ) {
- removeGapColumns();
- realignWithMafft();
- msa_prop = new MsaProperties( _msa, _report_aln_mean_identity );
+ _removed_seq_ids.add( id );
+ final Sequence deleted = _msa.deleteRow( id, true );
+ _removed_seqs.add( deleted );
+ removeGapColumns();
+ if ( isPrintMsaStatsWriteOutfileAndRealign( i ) || ( i == ( to_remove_ids.size() - 1 ) ) ) {
+ msa_prop = printMsaStatsWriteOutfileAndRealign( _realign, id );
msa_props.add( msa_prop );
- printMsaProperties( id, msa_prop );
- System.out.print( "(realigned)" );
System.out.println();
}
else if ( isPrintMsaStats( i ) ) {
- removeGapColumns();
msa_prop = new MsaProperties( _msa, _report_aln_mean_identity );
msa_props.add( msa_prop );
printMsaProperties( id, msa_prop );
System.out.println();
}
- ++i;
+ }
+ if ( _removed_seqs_out_base != null ) {
+ final String msg = writeAndAlignRemovedSeqs();
+ System.out.println();
+ System.out.println( msg );
}
return msa_props;
}
- private final boolean isPrintMsaStats( final int i ) {
- return ( ( _step_for_diagnostics < 2 ) || ( ( ( i + 1 ) % _step_for_diagnostics ) == 0 ) );
- }
-
- private final boolean isPrintMsaStatsWriteOutfileAndRealign( final int i ) {
- return ( ( _step < 2 ) || ( ( ( i + 1 ) % _step ) == 0 ) );
- }
-
public final void setGapRatio( final double gap_ratio ) {
_gap_ratio = gap_ratio;
}
+ public final void setMafftOptions( final String maffts_opts ) {
+ _maffts_opts = maffts_opts;
+ }
+
public final void setMinLength( final int min_length ) {
_min_length = min_length;
}
_realign = realign;
}
- public final void setStep( final int step ) {
- _step = step;
- }
-
- public final void setStepForDiagnostics( final int step_for_diagnostics ) {
- _step_for_diagnostics = step_for_diagnostics;
+ public final void setRemovedSeqsOutBase( final File removed_seqs_out_base ) {
+ _removed_seqs_out_base = removed_seqs_out_base;
}
public final void setReportAlnMeanIdentity( final boolean report_aln_mean_identity ) {
_report_aln_mean_identity = report_aln_mean_identity;
}
- final public String writeMsa( final File outfile, final MSA_FORMAT format, final String suffix ) throws IOException {
- final Double gr = MsaMethods.calcGapRatio( _msa );
- final String s = outfile + "_" + _msa.getNumberOfSequences() + "_" + _msa.getLength() + "_"
- + ForesterUtil.roundToInt( gr * 100 );
- writeMsa( _msa, s + suffix, format );
- return s;
+ public final void setStep( final int step ) {
+ _step = step;
+ }
+
+ public final void setStepForDiagnostics( final int step_for_diagnostics ) {
+ _step_for_diagnostics = step_for_diagnostics;
}
final public String writeAndAlignRemovedSeqs() throws IOException, InterruptedException {
final StringBuilder msg = new StringBuilder();
- final Msa removed = BasicMsa.createInstance( _removed_seqs );
- final String n = _removed_seqs_out_base + "_" + removed.getNumberOfSequences() + ".fasta";
- writeMsa( removed, n, MSA_FORMAT.FASTA );
- msg.append( "wrote " + removed.getNumberOfSequences() + " removed sequences to " + n );
+ final String n = _removed_seqs_out_base + "_" + _removed_seqs.size() + ".fasta";
+ SequenceWriter.writeSeqs( _removed_seqs, new File( n ), SEQ_FORMAT.FASTA, 100 );
+ msg.append( "wrote " + _removed_seqs.size() + " removed sequences to " + "\"" + n + "\"" );
if ( _realign ) {
final MsaInferrer mafft = Mafft.createInstance( _path_to_mafft );
final List<String> opts = new ArrayList<String>();
final Double gr = MsaMethods.calcGapRatio( removed_msa );
String s = _removed_seqs_out_base + "_" + removed_msa.getNumberOfSequences() + "_"
+ removed_msa.getLength() + "_" + ForesterUtil.roundToInt( gr * 100 );
- String suffix = "";
- if ( _output_format == MSA_FORMAT.FASTA ) {
- suffix = ".fasta";
- }
- else if ( _output_format == MSA_FORMAT.PHYLIP ) {
- suffix = ".aln";
- }
+ final String suffix = obtainSuffix();
s += suffix;
writeMsa( removed_msa, s, _output_format );
- msg.append( ", and as MSA of length " + removed_msa.getLength() + " to " + s );
+ msg.append( ", and as MSA of length " + removed_msa.getLength() + " to \"" + s + "\"" );
}
return msg.toString();
}
+ final public String writeMsa( final File outfile ) throws IOException {
+ final Double gr = MsaMethods.calcGapRatio( _msa );
+ final String s = outfile + "_" + _msa.getNumberOfSequences() + "_" + _msa.getLength() + "_"
+ + ForesterUtil.roundToInt( gr * 100 );
+ writeMsa( _msa, s + obtainSuffix(), _output_format );
+ return s;
+ }
+
final int calcNonGapResidues( final Sequence seq ) {
int ng = 0;
for( int i = 0; i < seq.getLength(); ++i ) {
return phy;
}
+ private final boolean isPrintMsaStats( final int i ) {
+ return ( ( ( _step == 1 ) && ( _step_for_diagnostics == 1 ) ) || ( ( _step_for_diagnostics > 0 ) && ( ( ( i + 1 ) % _step_for_diagnostics ) == 0 ) ) );
+ }
+
+ private final boolean isPrintMsaStatsWriteOutfileAndRealign( final int i ) {
+ return ( ( ( _step == 1 ) && ( _step_for_diagnostics == 1 ) ) || ( ( _step > 0 ) && ( ( ( i + 1 ) % _step ) == 0 ) ) );
+ }
+
+ private final StringBuilder msaPropertiesAsSB( final MsaProperties msa_properties ) {
+ final StringBuilder sb = new StringBuilder();
+ sb.append( msa_properties.getNumberOfSequences() );
+ sb.append( "\t" );
+ sb.append( msa_properties.getLength() );
+ sb.append( "\t" );
+ sb.append( NF_4.format( msa_properties.getGapRatio() ) );
+ if ( _report_aln_mean_identity ) {
+ sb.append( "\t" );
+ sb.append( NF_4.format( msa_properties.getAverageIdentityRatio() ) );
+ }
+ return sb;
+ }
+
+ private String obtainSuffix() {
+ if ( _output_format == MSA_FORMAT.FASTA ) {
+ return ".fasta";
+ }
+ else if ( _output_format == MSA_FORMAT.PHYLIP ) {
+ return ".aln";
+ }
+ return "";
+ }
+
private final Phylogeny pi( final String matrix ) {
final Phylogeny master_phy = inferNJphylogeny( PWD_DISTANCE_METHOD.KIMURA_DISTANCE, _msa, true, matrix );
final int seed = 15;
}
private final void printMsaProperties( final String id, final MsaProperties msa_properties ) {
- System.out.print( ForesterUtil.pad( _longest_id_length + 1, 20, ' ', false ) );
- System.out.print( "\t" );
- final StringBuilder sb = msaPropertiesAsSB( msa_properties );
- System.out.print( sb );
- System.out.print( "\t" );
- }
-
- private final StringBuilder msaPropertiesAsSB( final MsaProperties msa_properties ) {
- final StringBuilder sb = new StringBuilder();
- sb.append( msa_properties.getNumberOfSequences() );
- sb.append( "\t" );
- sb.append( msa_properties.getLength() );
- sb.append( "\t" );
- sb.append( NF_4.format( msa_properties.getGapRatio() ) );
- if ( _report_aln_mean_identity /*msa_properties.getAverageIdentityRatio() >= 0*/) {
- sb.append( "\t" );
- sb.append( NF_4.format( msa_properties.getAverageIdentityRatio() ) );
+ if ( ( _step == 1 ) || ( _step_for_diagnostics == 1 ) ) {
+ System.out.print( ForesterUtil.pad( id, _longest_id_length, ' ', false ) );
+ System.out.print( "\t" );
}
- return sb;
+ System.out.print( msaPropertiesAsSB( msa_properties ) );
+ System.out.print( "\t" );
}
- final private void printMsaStatsWriteOutfileAndRealign( final boolean realign, final String id )
+ final private MsaProperties printMsaStatsWriteOutfileAndRealign( final boolean realign, final String id )
throws IOException, InterruptedException {
if ( realign ) {
realignWithMafft();
printMsaProperties( id, msa_prop );
final String s = writeOutfile();
System.out.print( "-> " + s + ( realign ? "\t(realigned)" : "" ) );
+ return msa_prop;
+ }
+
+ private final void printTableHeader() {
+ if ( ( _step == 1 ) || ( _step_for_diagnostics == 1 ) ) {
+ System.out.print( ForesterUtil.pad( "Id", _longest_id_length, ' ', false ) );
+ System.out.print( "\t" );
+ }
+ System.out.print( "Seqs" );
+ System.out.print( "\t" );
+ System.out.print( "Length" );
+ System.out.print( "\t" );
+ System.out.print( "Gaps" );
+ System.out.print( "\t" );
+ if ( _report_aln_mean_identity ) {
+ System.out.print( "MSA qual" );
+ System.out.print( "\t" );
+ }
+ System.out.println();
}
final private void realignWithMafft() throws IOException, InterruptedException {
_msa.deleteGapOnlyColumns();
}
- final private static void writeMsa( final Msa msa, final String outfile, final MSA_FORMAT format )
- throws IOException {
- final Writer w = ForesterUtil.createBufferedWriter( outfile );
- msa.write( w, format );
- w.close();
- }
-
private final String writeOutfile() throws IOException {
- final String s = writeMsa( _out_file_base, MSA_FORMAT.PHYLIP, ".aln" );
- //writeMsa( _out_file_base, MSA_FORMAT.FASTA, ".fasta" );
+ final String s = writeMsa( _out_file_base );
return s;
}
return null;
}
- private final void printTableHeader() {
- System.out.print( ForesterUtil.pad( "Id", _longest_id_length + 1, ' ', false ) );
- System.out.print( "\t" );
- System.out.print( "Seqs" );
- System.out.print( "\t" );
- System.out.print( "Length" );
- System.out.print( "\t" );
- System.out.print( "Gaps" );
- System.out.print( "\t" );
- if ( _report_aln_mean_identity ) {
- System.out.print( "MSA qual" );
- System.out.print( "\t" );
- }
- System.out.println();
+ final private static void writeMsa( final Msa msa, final String outfile, final MSA_FORMAT format )
+ throws IOException {
+ final Writer w = ForesterUtil.createBufferedWriter( outfile );
+ msa.write( w, format );
+ w.close();
}
}
git://source.jalview.org / jalview.git / blobdiff