forester/java/src/org/forester/application/tap.java

   1
   2 package org.forester.application;
   3
   4 import java.io.BufferedWriter;
   5 import java.io.File;
   6 import java.io.FileInputStream;
   7 import java.io.IOException;
   8 import java.io.Writer;
   9 import java.util.ArrayList;
  10 import java.util.List;
  11
  12 import org.forester.io.parsers.FastaParser;
  13 import org.forester.io.parsers.GeneralMsaParser;
  14 import org.forester.io.writers.SequenceWriter;
  15 import org.forester.io.writers.SequenceWriter.SEQ_FORMAT;
  16 import org.forester.msa.BasicMsa;
  17 import org.forester.msa.Msa;
  18 import org.forester.msa.Msa.MSA_FORMAT;
  19 import org.forester.sequence.BasicSequence;
  20 import org.forester.sequence.MolecularSequence;
  21 import org.forester.util.CommandLineArguments;
  22 import org.forester.util.ForesterConstants;
  23 import org.forester.util.ForesterUtil;
  24
  25 public class tap {
  26
  27     final static private String PRG_NAME                = "tap";
  28     final static private String PRG_DATE                = "170327";
  29     final static private String PRG_DESC                = "Replacement of labels in multiple sequence files";
  30     final static private String PRG_VERSION             = "1.00";
  31     final static private String WWW                     = "https://sites.google.com/site/cmzmasek/home/software/forester";
  32     final static private String E_MAIL                  = "phyloxml@gmail.com";
  33     final static private String EXTRACT_TAXONOMY_OPTION = "t";
  34     final static private String ANNOTATION_OPTION       = "a";
  35     final static private String HELP_OPTION_1           = "help";
  36     final static private String HELP_OPTION_2           = "h";
  37
  38     public static void main( final String args[] ) {
  39         try {
  40             ForesterUtil.printProgramInformation( PRG_NAME,
  41                                                   PRG_DESC,
  42                                                   PRG_VERSION,
  43                                                   PRG_DATE,
  44                                                   E_MAIL,
  45                                                   WWW,
  46                                                   ForesterUtil.getForesterLibraryInformation() );
  47             CommandLineArguments cla = null;
  48             try {
  49                 cla = new CommandLineArguments( args );
  50             }
  51             catch ( final Exception e ) {
  52                 ForesterUtil.fatalError( PRG_NAME, e.getMessage() );
  53             }
  54             if ( cla.isOptionSet( HELP_OPTION_1 ) || cla.isOptionSet( HELP_OPTION_2 ) ) {
  55                 System.out.println();
  56                 print_help();
  57                 System.exit( 0 );
  58             }
  59             String input = null;
  60             String output = null;
  61             String list_file = null;
  62             String i = null;
  63             if ( args.length == 3 ) {
  64                 input = cla.getName( 0 );
  65                 output = cla.getName( 1 );
  66                 list_file = cla.getName( 2 );
  67             }
  68             else if ( args.length == 1 ) {
  69                 input = cla.getName( 0 );
  70                 i = null;
  71                 if ( input.toLowerCase().endsWith( ".fasta" ) ) {
  72                     i = input.substring( 0, input.length() - 7 );
  73                 }
  74                 else if ( input.toLowerCase().endsWith( ".fsa" ) ) {
  75                     i = input.substring( 0, input.length() - 5 );
  76                 }
  77                 else {
  78                     i = input;
  79                 }
  80                 output = i + ForesterConstants.ID_NORMALIZED_FASTA_FILE_SUFFIX;
  81                 list_file = i + ForesterConstants.ID_MAP_FILE_SUFFIX;
  82             }
  83             else {
  84                 print_help();
  85                 System.exit( -1 );
  86             }
  87             final List<String> allowed_options = new ArrayList<>();
  88             allowed_options.add( ANNOTATION_OPTION );
  89             final String dissallowed_options = cla.validateAllowedOptionsAsString( allowed_options );
  90             if ( dissallowed_options.length() > 0 ) {
  91                 ForesterUtil.fatalError( PRG_NAME, "unknown option(s): " + dissallowed_options );
  92             }
  93             final File outfile_file = new File( output );
  94             final File listfile = new File( list_file );
  95             final File input_file = new File( input );
  96             final String error1 = ForesterUtil.isWritableFile( outfile_file );
  97             if ( !ForesterUtil.isEmpty( error1 ) ) {
  98                 ForesterUtil.fatalError( PRG_NAME, error1 );
  99             }
 100             final String error2 = ForesterUtil.isWritableFile( listfile );
 101             if ( !ForesterUtil.isEmpty( error2 ) ) {
 102                 ForesterUtil.fatalError( PRG_NAME, error2 );
 103             }
 104             final String error3 = ForesterUtil.isReadableFile( input_file );
 105             if ( !ForesterUtil.isEmpty( error3 ) ) {
 106                 ForesterUtil.fatalError( PRG_NAME, error3 );
 107             }
 108             final boolean fasta_like = ForesterUtil.isLooksLikeFasta( input_file );
 109             final Msa.MSA_FORMAT output_format = MSA_FORMAT.FASTA;
 110             System.out.println();
 111             System.out.println( "Input alignment       : " + input );
 112             System.out.println( "Output alignment      : " + output );
 113             System.out.println( "Name list             : " + list_file );
 114             if ( fasta_like ) {
 115                 System.out.println( "Input format          : Fasta" );
 116             }
 117             else {
 118                 System.out.println( "Input format          : Phylip like" );
 119             }
 120             if ( output_format == MSA_FORMAT.FASTA ) {
 121                 System.out.println( "Output format         : Fasta" );
 122             }
 123             else if ( output_format == MSA_FORMAT.NEXUS ) {
 124                 System.out.println( "Output format         : Nexus" );
 125             }
 126             else if ( output_format == MSA_FORMAT.PHYLIP ) {
 127                 System.out.println( "Output format         : Phylip" );
 128             }
 129             System.out.println();
 130
 131             final List<MolecularSequence> seqs;
 132             final FileInputStream is = new FileInputStream( input_file );
 133             if ( FastaParser.isLikelyFasta( input_file ) ) {
 134                 seqs = FastaParser.parse( is );
 135             }
 136             else {
 137                 seqs = GeneralMsaParser.parseSeqs( is );
 138             }
 139             if ( seqs == null ) {
 140                 ForesterUtil.fatalError( PRG_NAME, "failed to read MSA" );
 141             }
 142             if ( seqs.size() < 1 ) {
 143                 ForesterUtil.fatalError( PRG_NAME, "MSA seems to be devoid of sequences" );
 144             }
 145            // TODO print number of seqs
 146            // TODO print number min length
 147            // TODO print max length
 148            // TODO OR
 149           //  TODO print length is aligned
 150           //  TODO if no aligned no phylip or nexus outpt
 151             //
 152
 153             final List<MolecularSequence> seqs2 = new ArrayList<>();
 154             int counter = 0;
 155             final BufferedWriter writer = ForesterUtil.createBufferedWriter( list_file );
 156             for( final MolecularSequence seq : seqs ) {
 157                 final String new_name = modify_name( seq.getIdentifier(), counter++, writer );
 158                 final MolecularSequence ns = BasicSequence.createSequence( new_name,
 159                                                                            seq.getMolecularSequenceAsString() );
 160                 seqs2.add( ns );
 161             }
 162             writer.flush();
 163             writer.close();
 164             final BufferedWriter seq_writer = ForesterUtil.createBufferedWriter( outfile_file );
 165             if ( ( output_format == MSA_FORMAT.NEXUS ) || ( output_format == MSA_FORMAT.PHYLIP ) ) {
 166                 final Msa m = BasicMsa.createInstance( seqs2 );
 167                 m.write( seq_writer, output_format );
 168             }
 169             else if ( output_format == MSA_FORMAT.FASTA ) {
 170                 SequenceWriter.writeSeqs( seqs2, seq_writer, SEQ_FORMAT.FASTA, 60 );
 171             }
 172             seq_writer.flush();
 173             seq_writer.close();
 174             //                    Util.print_message( PRG_NAME, "wrote: " + list_file )
 175             //                    Util.print_message( PRG_NAME, "wrote: " + output )
 176         }
 177         catch ( final IllegalArgumentException e ) {
 178             ForesterUtil.fatalError( PRG_NAME, e.getMessage() );
 179         }
 180         catch ( final Exception e ) {
 181             e.printStackTrace();
 182             ForesterUtil.fatalError( PRG_NAME, "Unexpected errror!" );
 183         }
 184     }
 185
 186     final static String modify_name( final String desc, final int counter, final Writer writer ) throws IOException {
 187         desc.replaceAll( "\\s+", " " );
 188         final String new_desc = Integer.toHexString( counter );
 189         if ( new_desc.length() > 9 ) {
 190             ForesterUtil.fatalError( PRG_NAME,
 191                                      "shortened identifier [" + new_desc + "] is too long (" + new_desc.length()
 192                                              + " characters)" );
 193         }
 194         writer.write( new_desc + "\t" + desc + "\n" );
 195         return new_desc;
 196     }
 197
 198     private final static void print_help() {
 199         System.out.println( "Usage:" );
 200         System.out.println();
 201         System.out.println( PRG_NAME + " [options] <gene tree file> <query>" );
 202         System.out.println();
 203         System.out.println( " options:" );
 204         //System.out.println( "  -" + SEP_OPTION + "=<separator>: the separator to be used" );
 205         System.out.println();
 206         System.out.println( "Example:" );
 207         System.out.println();
 208         System.out.println( " " + PRG_NAME + " -s=. my_tree.xml A.1.1.1" );
 209         System.out.println();
 210     }
 211 }