import org.forester.io.parsers.FastaParser;
import org.forester.io.parsers.GeneralMsaParser;
import org.forester.msa.Msa;
-import org.forester.msa.MsaCompactor;
-import org.forester.msa.MsaMethods;
+import org.forester.msa.MsaInferrer;
+import org.forester.msa_compactor.MsaCompactor;
import org.forester.util.CommandLineArguments;
import org.forester.util.ForesterUtil;
public class msa_compactor {
- final static private String HELP_OPTION_1 = "help";
- final static private String HELP_OPTION_2 = "h";
- final static private String REMOVE_WORST_OFFENDERS_OPTION = "w";
- final static private String AV_GAPINESS_OPTION = "a";
- final static private String LENGTH_OPTION = "l";
- final static private String REALIGN_OPTION = "r";
- final static private String PRG_NAME = "msa_compactor";
- final static private String PRG_DESC = "multiple sequnce aligment compactor";
- final static private String PRG_VERSION = "0.90";
- final static private String PRG_DATE = "2012.07.11";
- final static private String E_MAIL = "phylosoft@gmail.com";
- final static private String WWW = "www.phylosoft.org/forester/";
+ final static private String HELP_OPTION_1 = "help";
+ final static private String HELP_OPTION_2 = "h";
+ final static private String REMOVE_WORST_OFFENDERS_OPTION = "r";
+ final static private String AV_GAPINESS_OPTION = "g";
+ final static private String STEP_OPTION = "s";
+ final static private String LENGTH_OPTION = "l";
+ final static private String REALIGN_OPTION = "a";
+ final static private String PATH_TO_MAFFT_OPTION = "mafft";
+ final static private String DO_NOT_NORMALIZE_FOR_EFF_LENGTH_OPTION = "nn";
+ final static private String PRG_NAME = "msa_compactor";
+ final static private String PRG_DESC = "multiple sequnce aligment compactor";
+ final static private String PRG_VERSION = "0.01";
+ final static private String PRG_DATE = "140314";
+ final static private String E_MAIL = "phylosoft@gmail.com";
+ final static private String WWW = "https://sites.google.com/site/cmzmasek/home/software/forester";
public static void main( final String args[] ) {
try {
final CommandLineArguments cla = new CommandLineArguments( args );
- if ( cla.isOptionSet( HELP_OPTION_1 ) || cla.isOptionSet( HELP_OPTION_2 ) ) {
+ if ( cla.isOptionSet( HELP_OPTION_1 ) || cla.isOptionSet( HELP_OPTION_2 ) || ( cla.getNumberOfNames() != 2 ) ) {
printHelp();
System.exit( 0 );
}
final File in = cla.getFile( 0 );
+ final File out = cla.getFile( 1 );
int worst_remove = -1;
double av = -1;
int length = -1;
- final int step = 5;
+ int step = 1;
boolean realign = false;
- // int to = 0;
- // int window = 0;
- // int step = 0;
+ boolean norm = true;
+ String path_to_mafft = null;
final List<String> allowed_options = new ArrayList<String>();
allowed_options.add( REMOVE_WORST_OFFENDERS_OPTION );
allowed_options.add( AV_GAPINESS_OPTION );
allowed_options.add( LENGTH_OPTION );
allowed_options.add( REALIGN_OPTION );
+ allowed_options.add( DO_NOT_NORMALIZE_FOR_EFF_LENGTH_OPTION );
+ allowed_options.add( STEP_OPTION );
+ allowed_options.add( PATH_TO_MAFFT_OPTION );
final String dissallowed_options = cla.validateAllowedOptionsAsString( allowed_options );
if ( dissallowed_options.length() > 0 ) {
ForesterUtil.fatalError( PRG_NAME, "unknown option(s): " + dissallowed_options );
if ( cla.isOptionSet( LENGTH_OPTION ) ) {
length = cla.getOptionValueAsInt( LENGTH_OPTION );
}
+ if ( cla.isOptionSet( STEP_OPTION ) ) {
+ step = cla.getOptionValueAsInt( STEP_OPTION );
+ }
if ( cla.isOptionSet( REALIGN_OPTION ) ) {
realign = true;
}
+ if ( cla.isOptionSet( PATH_TO_MAFFT_OPTION ) ) {
+ if ( !realign ) {
+ ForesterUtil.fatalError( PRG_NAME, "no need to indicate path to MAFFT without realigning" );
+ }
+ path_to_mafft = cla.getOptionValueAsCleanString( PATH_TO_MAFFT_OPTION );
+ }
+ if ( cla.isOptionSet( DO_NOT_NORMALIZE_FOR_EFF_LENGTH_OPTION ) ) {
+ norm = false;
+ }
// else if ( cla.isOptionSet( STEP_OPTION ) && cla.isOptionSet( WINDOW_OPTION ) ) {
// step = cla.getOptionValueAsInt( STEP_OPTION );
// window = cla.getOptionValueAsInt( WINDOW_OPTION );
// printHelp();
// System.exit( 0 );
// }
+ if ( realign ) {
+ if ( ForesterUtil.isEmpty( path_to_mafft ) ) {
+ path_to_mafft = MsaCompactor.guessPathToMafft();
+ }
+ checkPathToMafft( path_to_mafft );
+ }
Msa msa = null;
final FileInputStream is = new FileInputStream( in );
if ( FastaParser.isLikelyFasta( in ) ) {
else {
msa = GeneralMsaParser.parse( is );
}
- System.out.println( msa.toString() );
- System.out.println( MsaMethods.calcBasicGapinessStatistics( msa ).arithmeticMean() );
MsaCompactor mc = null;
if ( worst_remove > 0 ) {
- mc = MsaCompactor.removeWorstOffenders( msa, worst_remove, realign );
+ mc = MsaCompactor.removeWorstOffenders( msa, worst_remove, step, realign, norm, path_to_mafft, out );
}
else if ( av > 0 ) {
- mc = MsaCompactor.reduceGapAverage( msa, av, step, realign );
+ mc = MsaCompactor.reduceGapAverage( msa, av, step, realign, norm, path_to_mafft, out );
}
else if ( length > 0 ) {
- mc = MsaCompactor.reduceLength( msa, length, step, realign );
- }
- System.out.println( mc.getMsa().toString() );
- System.out.println( MsaMethods.calcBasicGapinessStatistics( mc.getMsa() ).arithmeticMean() );
- for( final String id : mc.getRemovedSeqIds() ) {
- System.out.println( id );
+ if ( length >= msa.getLength() ) {
+ ForesterUtil.fatalError( PRG_NAME, "target MSA length (" + length
+ + ") is greater than or equal to MSA original length (" + msa.getLength() + ")" );
+ }
+ // TODO if < shortest seq -> error
+ mc = MsaCompactor.reduceLength( msa, length, step, realign, norm, path_to_mafft, out );
}
+ //System.out.println( MsaMethods.calcGapRatio( mc.getMsa() ) );
+ // for( final String id : mc.getRemovedSeqIds() ) {
+ // System.out.println( id );
+ //}
+ //mc.writeMsa( out, MSA_FORMAT.PHYLIP, ".aln" );
}
catch ( final Exception e ) {
e.printStackTrace();
}
}
+ private static void checkPathToMafft( final String path_to_mafft ) {
+ if ( !ForesterUtil.isEmpty( path_to_mafft ) && MsaInferrer.isInstalled( path_to_mafft ) ) {
+ ForesterUtil.programMessage( PRG_NAME, "using MAFFT at \"" + path_to_mafft + "\"" );
+ }
+ else {
+ if ( ForesterUtil.isEmpty( path_to_mafft ) ) {
+ ForesterUtil.fatalError( PRG_NAME, "no MAFFT executable found, use -\"" + PATH_TO_MAFFT_OPTION
+ + "=<path to MAFFT>\" option" );
+ }
+ else {
+ ForesterUtil.fatalError( PRG_NAME, "no MAFFT executable at \"" + path_to_mafft + "\"" );
+ }
+ }
+ }
+
private static void printHelp() {
ForesterUtil.printProgramInformation( PRG_NAME,
PRG_DESC,
E_MAIL,
WWW,
ForesterUtil.getForesterLibraryInformation() );
+ final String path_to_mafft = MsaCompactor.guessPathToMafft();
+ String mafft_comment;
+ if ( !ForesterUtil.isEmpty( path_to_mafft ) ) {
+ mafft_comment = " (using " + path_to_mafft + ")";
+ }
+ else {
+ mafft_comment = " (no path to MAFFT found, use -\"" + PATH_TO_MAFFT_OPTION + "=<path to MAFFT>\" option";
+ }
System.out.println( "Usage:" );
System.out.println();
- System.out.println( PRG_NAME + " <options> <msa input file>" );
+ System.out.println( PRG_NAME + " <options> <msa input file> <output file>" );
System.out.println();
System.out.println( " options: " );
System.out.println();
- // System.out.println( " -" + FROM_OPTION + "=<integer>: from (msa column)" );
- // System.out.println( " -" + TO_OPTION + "=<integer>: to (msa column)" );
- // System.out.println( " or" );
- // System.out.println( " -" + WINDOW_OPTION + "=<integer>: window size (msa columns)" );
- System.out.println( " -" + REMOVE_WORST_OFFENDERS_OPTION + "=<integer>: step size (msa columns)" );
+ System.out.println( " -" + REMOVE_WORST_OFFENDERS_OPTION
+ + "=<integer> number of worst offender sequences to remove" );
+ System.out.println( " -" + LENGTH_OPTION + "=<integer> target MSA length" );
+ System.out.println( " -" + AV_GAPINESS_OPTION + "=<decimal> gap %" );
+ System.out.println( " -" + STEP_OPTION + "=<decimal> step" );
+ System.out.println( " -" + REALIGN_OPTION + " to realign using MAFFT" + mafft_comment );
System.out.println();
System.out.println();
System.out.println();
git://source.jalview.org / jalview.git / blobdiff