src/jalview/io/RnamlFile.java

   1 /*
   2  * Jalview - A Sequence Alignment Editor and Viewer (Version 2.8)
   3  * Copyright (C) 2012 J Procter, AM Waterhouse, LM Lui, J Engelhardt, G Barton, M Clamp, S Searle
   4  *
   5  * This file is part of Jalview.
   6  *
   7  * Jalview is free software: you can redistribute it and/or
   8  * modify it under the terms of the GNU General Public License
   9  * as published by the Free Software Foundation, either version 3 of the License, or (at your option) any later version.
  10  *
  11  * Jalview is distributed in the hope that it will be useful, but
  12  * WITHOUT ANY WARRANTY; without even the implied warranty
  13  * of MERCHANTABILITY or FITNESS FOR A PARTICULAR
  14  * PURPOSE.  See the GNU General Public License for more details.
  15  *
  16  * You should have received a copy of the GNU General Public License along with Jalview.  If not, see <http://www.gnu.org/licenses/>.
  17  */
  18 package jalview.io;
  19
  20 import jalview.datamodel.AlignmentAnnotation;
  21 import jalview.datamodel.Annotation;
  22 import jalview.datamodel.SecondaryStructureAnnotation;
  23 import jalview.datamodel.Sequence;
  24 import jalview.datamodel.SequenceFeature;
  25 import jalview.datamodel.SequenceI;
  26 import jalview.ws.jws1.Annotate3D;
  27
  28 import java.io.BufferedReader;
  29 import java.io.FileNotFoundException;
  30 import java.io.IOException;
  31 import java.io.Reader;
  32 import java.io.FileReader;
  33 import java.lang.reflect.Array;
  34 import java.util.ArrayList;
  35 import java.util.Collections;
  36 import java.util.Hashtable;
  37 import java.util.List;
  38 import java.util.Vector;
  39
  40 import javax.xml.parsers.ParserConfigurationException;
  41 import javax.xml.parsers.SAXParser;
  42 import javax.xml.parsers.SAXParserFactory;
  43
  44 import org.xml.sax.InputSource;
  45 import org.xml.sax.SAXException;
  46
  47 import com.stevesoft.pat.Regex;
  48
  49 import fr.orsay.lri.varna.exceptions.ExceptionFileFormatOrSyntax;
  50 import fr.orsay.lri.varna.exceptions.ExceptionLoadingFailed;
  51 import fr.orsay.lri.varna.exceptions.ExceptionPermissionDenied;
  52 import fr.orsay.lri.varna.exceptions.ExceptionUnmatchedClosingParentheses;
  53 import fr.orsay.lri.varna.factories.RNAFactory;
  54 import fr.orsay.lri.varna.factories.RNAFactory.RNAFileType;
  55 import fr.orsay.lri.varna.models.rna.RNA;
  56 import fr.orsay.lri.varna.utils.RNAMLParser;
  57
  58
  59
  60
  61 public class RnamlFile extends AlignFile
  62 {
  63          public String id;
  64          protected ArrayList<RNA> result;
  65
  66          public RnamlFile()
  67           {
  68                  super();
  69
  70           }
  71
  72          public RnamlFile(String inFile, String type) throws IOException, ExceptionFileFormatOrSyntax, ParserConfigurationException, SAXException, ExceptionPermissionDenied, ExceptionLoadingFailed, InterruptedException, ExceptionUnmatchedClosingParentheses
  73           {
  74             super(inFile, type);
  75
  76           }
  77
  78           public RnamlFile(FileParse source) throws IOException, ExceptionFileFormatOrSyntax, ParserConfigurationException, SAXException, ExceptionPermissionDenied, ExceptionLoadingFailed, InterruptedException, ExceptionUnmatchedClosingParentheses
  79           {
  80             super(source);
  81
  82           }
  83
  84 //        public RnamlFile(BufferedReader r) throws IOException, ExceptionFileFormatOrSyntax, ParserConfigurationException, SAXException, ExceptionPermissionDenied, ExceptionLoadingFailed
  85 //        {
  86 //                      super();
  87 //                  parse(r);
  88 //                  // sets the index of each sequence in the alignment
  89 //                  for( int i=0,c=seqs.size(); i<c; i++ ) {
  90 //                      seqs.get(i).setIndex(i);
  91 //                  }
  92 //
  93 //
  94 //        }
  95
  96         public BufferedReader CreateReader() throws FileNotFoundException
  97         {
  98                  FileReader fr = null;
  99                   fr = new FileReader(inFile);
 100
 101                   BufferedReader r = new BufferedReader (fr);
 102                   return r;
 103         }
 104
 105
 106         @SuppressWarnings("unchecked")
 107         public void parse() throws FileNotFoundException, ExceptionPermissionDenied, ExceptionLoadingFailed, ExceptionFileFormatOrSyntax
 108         {
 109                  result = RNAFactory.loadSecStrRNAML(getReader());
 110
 111                  System.out.println("this is the secondary scructure:" +result.size());
 112                   //System.out.println("this is the secondary scructure:" +result.toString());
 113                   //System.out.println("this is the secondary scructure:" +result.size());
 114                  SequenceI[] seqs = new SequenceI[result.size()];
 115                   System.out.println(type); //the type is "File"
 116                   System.out.println(inFile );//inFile is the path
 117
 118                  for(int i=0;i<result.size();i++)
 119                         {
 120                          RNA current = result.get(i);
 121                   System.out.println(current.getSeq());
 122                   //System.out.println(result.get(i).getStructBPSEQ());
 123                   System.out.println(result.get(i).getStructDBN(true));
 124                   System.out.println(i);
 125                   String rna =current.getStructDBN(true);
 126                   //System.out.println(result.get(0).getAnnotations());
 127                   String seq = current.getSeq();
 128
 129                   int begin=1;
 130                   int end = seq.length();
 131
 132                   id = current.getName();
 133                   seqs[i] = new Sequence(id, seq, begin, end);
 134                   //System.out.println(seq);
 135                   System.out.println("Rna is still "+rna);
 136                   //SequenceFeature feat = new SequenceFeature(type, "",5, 13, 0f, null);
 137                   //seqs[i].addSequenceFeature(feat);
 138
 139                   String[] annot=new String[rna.length()] ;
 140                   Annotation[] ann = new Annotation[rna.length()];
 141
 142                   for(int j=0;j<rna.length();j++)
 143                         {
 144                           annot[j] =rna.substring(j,j+1);
 145
 146                         }
 147
 148                   for(int k=0;k<rna.length();k++)
 149                 {
 150                           ann[k] = new Annotation(annot[k], "", jalview.schemes.ResidueProperties.getRNASecStrucState(annot[k]).charAt(0), 0f);
 151
 152                 }
 153                   AlignmentAnnotation align = new AlignmentAnnotation("Sec. str.",current.getID(),ann);
 154
 155                   seqs[i].addAlignmentAnnotation(align);
 156                   seqs[i].setRNA(result.get(i));
 157                   this.annotations.addElement(align);
 158                 }
 159           this.setSeqs(seqs);
 160
 161
 162
 163
 164         }
 165
 166
 167           public void parse(BufferedReader r) throws ExceptionPermissionDenied, ExceptionLoadingFailed, ExceptionFileFormatOrSyntax
 168           {
 169                    System.out.println("parse with Buffer (RnamlFile");
 170                   Hashtable seqhash = new Hashtable();
 171                   Vector headers = new Vector();
 172                   //ArrayList<String> seq =new ArrayList();
 173                   //System.out.println(r);
 174
 175                   result = RNAFactory.loadSecStr(r,RNAFileType.FILE_TYPE_RNAML);
 176
 177                   System.out.println("this is the secondary scructure:" +result.toArray());
 178                   System.out.println("this is the secondary scructure:" +result.toString());
 179                   System.out.println("this is the secondary scructure:" +result.size());
 180                         for(int i=0;i<result.size();i++)
 181                         {
 182                   System.out.println(result.get(0).getSeq());
 183                   System.out.println(result.get(0).getStructBPSEQ());
 184                   System.out.println(result.get(0).getStructDBN(true));
 185                   String rna1 =result.get(0).getStructDBN(true);
 186                   System.out.println(result.get(0).getAnnotations());
 187                   String seq = result.get(0).getSeq();
 188                   System.out.println(type); //the type is "File"
 189                   System.out.println(inFile );//inFile is the path
 190                   int begin=0;
 191                   int end = seq.length()-1;
 192                   SequenceI[] seqs = new SequenceI[1];
 193                   id = safeName(getDataName());
 194                   seqs[0] = new Sequence(id, seq, begin, end);
 195                   SequenceFeature feat = new SequenceFeature(type, "",5, 13, 0f, null);
 196                   seqs[0].addSequenceFeature(feat);
 197                   this.setSeqs(seqs);
 198
 199                   String[] annot=new String[10000] ;
 200                   Annotation[] ann = new Annotation[1000];
 201                   for(int j=0;j<rna1.length();j++)
 202                         {
 203                   annot[j] =rna1.substring(j,j+1);
 204                   //System.out.println(annot[j]);
 205                         }
 206
 207                   for(int k=0;k<rna1.length();k++)
 208                 {
 209                           ann[k] = new Annotation(annot[k], "", ' ', 0f);
 210
 211                           ann[k].secondaryStructure = jalview.schemes.ResidueProperties.getRNASecStrucState(annot[k]).charAt(0);
 212
 213                 }
 214                   AlignmentAnnotation align = new AlignmentAnnotation("Secondary structure","",ann);
 215                   seqs[0].addAlignmentAnnotation(align);
 216                   this.annotations.addElement(align);
 217                 }
 218                 //Annotate3D tintin = new Annotate3D();
 219
 220                 //System.out.println(tintin);
 221
 222
 223
 224
 225
 226
 227
 228
 229 //                                      for(RNAMLParser.RNATmp rnaTmp : RNAMLData.getMolecules())
 230 //                                      {
 231 //                                              // Retrieving parsed data
 232 //                                              seq = rnaTmp.getSequence(); //contient la séquence sous forme de liste
 233 //                                              System.out.println(seq);
 234 //                                              String seqAsStr = "";
 235 //                                              for(String s:seq)
 236 //                                              {
 237 //                                                      // Only consider last character, e.g. in case of modified nucleotide
 238 //                                                      seqAsStr += s.charAt(s.length()-1);
 239 //                                              }
 240 //                                              int begin=0;
 241 //                                              int end = seqAsStr.length()-1;
 242 //                                              SequenceI[] seqs = new SequenceI[1];
 243 //                                              id = safeName(getDataName());
 244 //                                              seqs[0] = new Sequence(id, seqAsStr, begin, end);
 245 //                                              SequenceFeature feat = new SequenceFeature(type, "",5, 13, 0f, null);
 246 //
 247 //                          seqs[0].addSequenceFeature(feat);
 248 //                                              this.setSeqs(seqs);
 249
 250                                                 //String pos = annot.substring(1, 2);
 251                                                 //System.out.println(annot);
 252                                                 //System.out.println(pos);
 253
 254                                            // ann[1] = new Annotation(pos, "", ' ', 0f);
 255                                            // ann[1].secondaryStructure = 'A';
 256                                             //System.out.println(ann[1]);
 257
 258                                            // toto.annotations=ann;
 259
 260
 261
 262 ////                                             Annotation[] els = new Annotation[seq.size()];
 263 //
 264 //                                              //System.err.println(""+seq.size());
 265 //                                          Vector<RNAMLParser.BPTemp> allbpsTmp = rnaTmp.getStructure();
 266 //                                          String[] annot = new String[10000];
 267 //
 268 //                                              //Annotation[] ann = new Annotation[allbpsTmp.size()];
 269 //                                              for (int i = 0; i < allbpsTmp.size(); i++) {
 270 //                                              RNAMLParser.BPTemp bp = allbpsTmp.get(i);
 271 //                                              //System.err.println("toto ="+bp);
 272 //                                              int bp5 = bp.pos5;
 273 //                                              annot[bp5]= "(";
 274 //                                              //annot[bp5]= '(';
 275 //                                              System.out.println(annot[bp5]);
 276 //                                              //System.out.println("bp5 ="+bp5);
 277 //                                              int bp3 = bp.pos3;
 278 //
 279 //                                              annot[bp3]= ")";
 280 //                                              System.out.println(annot[bp3]);
 281 //                                              //System.out.println(annot.get(bp3));
 282 //
 283 //                                              //String pos1 = annot.substring(bp5, bp5+1);
 284 //                                              //String pos2 = annot.substring(bp3, bp3+1);
 285 //                                              //ann[i] = new Annotation(pos1, "", ' ', 0f);
 286 //                                              //ann[i+1] = new Annotation(pos2, "", ' ', 0f);
 287 //                                              //System.out.println(pos1);
 288 //
 289 //                                              //System.out.println("bp3 ="+bp3);
 290 //                                              //ModeleBP newStyle = bp.createBPStyle(mb, part);
 291 //                                              }
 292 //                                              Annotation[] ann = new Annotation[annot.length+1];
 293 //                                              for(int i=0;i<annot.length;i++)
 294 //                                              {
 295 //                                                      if(annot[i]==null)
 296 //                                                      {
 297 //                                                              annot[i]=".";
 298 //                                                      }
 299 //
 300 //                                                      ann[i] = new Annotation(annot[i], "", ' ', 0f);
 301 //                                              //System.out.print(annot[i]);
 302 //                                              }
 303 //
 304 //                                              AlignmentAnnotation toto = new AlignmentAnnotation("Secondary structure","",ann);
 305 //                                              this.annotations.addElement(toto);
 306
 307                                                 // Creating empty structure of suitable size
 308 //                                              int[] str = new int[seq.size()];
 309 //                                              for (int i=0;i<str.length;i++)
 310 //                                              { str[i] = -1; }
 311 //                                              Vector<RNAMLParser.BPTemp> allbpsTmp = rnaTmp.getStructure();
 312 //                                              for (int i = 0; i < allbpsTmp.size(); i++) {
 313 //                                                      RNAMLParser.BPTemp bp = allbpsTmp.get(i);
 314 //                                                      //System.err.println("toto ="+bp);
 315 //                                                      int bp5 = bp.pos5;
 316 //                                                      //System.out.println("bp5 ="+bp5);
 317 //                                                      int bp3 = bp.pos3;
 318                                                         //System.out.println("bp3 ="+bp3);
 319                                                         //ModeleBP newStyle = bp.createBPStyle(mb, part);
 320 //                                              }
 321                                                 //Annotation ann;
 322                                             //ann = new Annotation(bp5, "", ' ', 0f);
 323
 324                                         //}
 325
 326
 327
 328 //                              return result;
 329
 330 }
 331
 332
 333           public static String print(SequenceI[] s)
 334           {
 335             return "not yet implemented";
 336           }
 337
 338           public String print()
 339           {
 340                 System.out.print("affiche :");
 341             return print(getSeqsAsArray());
 342           }
 343
 344           /**
 345            * make a friendly ID string.
 346            *
 347            * @param dataName
 348            * @return truncated dataName to after last '/'
 349            */
 350           private String safeName(String dataName)
 351           {
 352             int b = 0;
 353             while ((b = dataName.indexOf("/")) > -1 && b < dataName.length() )
 354             {
 355                 dataName = dataName.substring(b + 1).trim();
 356
 357             }
 358             int e = (dataName.length() - dataName.indexOf("."))+1;
 359             dataName = dataName.substring(1,e).trim();
 360             return dataName;
 361           }
 362
 363           public ArrayList getRNA()
 364           {
 365                   return result;
 366           }
 367
 368           public Vector getAnnot()
 369           {
 370                   return annotations;
 371           }
 372 }