src/org/vamsas/objects/utils/SeqAln.java

   1 /*
   2  * Created on 17-May-2005
   3  *
   4  * TODO To change the template for this generated file go to
   5  * Window - Preferences - Java - Code Style - Code Templates
   6  */
   7 package org.vamsas.objects.utils;
   8
   9 import java.io.BufferedOutputStream;
  10 import java.io.BufferedReader;
  11 import java.io.BufferedWriter;
  12 import java.io.IOException;
  13 import java.io.InputStream;
  14 import java.io.InputStreamReader;
  15 import java.util.Hashtable;
  16 import java.util.StringTokenizer;
  17 import java.util.Vector;
  18 import java.util.regex.Pattern;
  19
  20 import vamsas.objects.simple.Alignment;
  21 import vamsas.objects.simple.Sequence;
  22 import vamsas.objects.simple.SequenceSet;
  23
  24 /**
  25  * @author jimp
  26  *
  27  * TODO To change the template for this generated type comment go to
  28  * Window - Preferences - Java - Code Style - Code Templates
  29  */
  30 public class SeqAln extends vamsas.objects.simple.Alignment {
  31
  32         public static Sequence[] ReadClustalFile(InputStream os) throws Exception {
  33
  34                 Pattern nonGap = Pattern.compile("[A-Z*0-9]", Pattern.CASE_INSENSITIVE);
  35                 String gapchars = "";
  36                 char gapchar='-';
  37
  38                 int     i    = 0;
  39             boolean flag = false;
  40
  41             Vector    headers = new Vector();
  42             Hashtable seqhash = new Hashtable();
  43             Sequence[] seqs=null;
  44             int noSeqs = 0;
  45             String line;
  46
  47             try {
  48                 BufferedReader ins = new BufferedReader(new InputStreamReader(os));
  49                 while ((line = ins.readLine()) != null) {
  50                         if (line.indexOf(" ") != 0) {
  51                                 java.util.StringTokenizer str = new StringTokenizer(line," ");
  52                                 String id = "";
  53
  54                                 if (str.hasMoreTokens()) {
  55                                         id = str.nextToken();
  56                                         if (id.equals("CLUSTAL")) {
  57                                                 flag = true;
  58                                         } else {
  59                                                 if (flag) {
  60                                                         StringBuffer tempseq;
  61                                                         if (seqhash.containsKey(id)) {
  62                                                                 tempseq = (StringBuffer)seqhash.get(id);
  63                                                         } else {
  64                                                                 tempseq = new StringBuffer();
  65                                                                 seqhash.put(id,tempseq);
  66                                                         }
  67
  68                                                         if (!(headers.contains(id))) {
  69                                                                 headers.addElement(id);
  70                                                         }
  71
  72                                                         tempseq.append(str.nextToken());
  73                                                 }
  74                                         }
  75                                 }
  76                         }
  77                 }
  78
  79             } catch (IOException e) {
  80                 throw(new Exception("Exception parsing clustal file ",e));
  81             }
  82
  83             if (flag) {
  84                 noSeqs = headers.size();
  85
  86                 //Add sequences to the hash
  87                 seqs = new Sequence[headers.size()];
  88               for (i = 0; i < headers.size(); i++ ) {
  89                 if ( seqhash.get(headers.elementAt(i)) != null) {
  90
  91                         Sequence newSeq = new Sequence(headers.elementAt(i).toString(),
  92                                                  seqhash.get(headers.elementAt(i).toString()).toString());
  93
  94                         seqs[i]=newSeq;
  95
  96                 } else {
  97                   throw(new Exception("Bizarreness! Can't find sequence for " + headers.elementAt(i)));
  98                 }
  99               }
 100             }
 101             return seqs;
 102           }
 103
 104           public static void WriteClustalWAlignment(java.io.OutputStream os, Alignment seqAl) throws IOException {
 105                 Sequence[] s = seqAl.getSeqs().getSeqs();
 106
 107                 java.io.BufferedWriter out = new BufferedWriter(new java.io.OutputStreamWriter(os));
 108
 109                 out.write("CLUSTAL\n\n");
 110
 111             int max = 0;
 112             int maxid = 0;
 113
 114             int i = 0;
 115
 116             while (i < s.length && s[i] != null) {
 117               String tmp = s[i].getId();
 118
 119               if (s[i].getSeq().length() > max) {
 120                 max = s[i].getSeq().length();
 121               }
 122               if (tmp.length() > maxid) {
 123                 maxid = tmp.length();
 124               }
 125               i++;
 126             }
 127
 128             if (maxid < 15) {
 129               maxid = 15;
 130             }
 131             maxid++;
 132             int len = 60;
 133             int nochunks =  max / len + 1;
 134
 135             for (i = 0; i < nochunks; i++) {
 136               int j = 0;
 137               while ( j < s.length && s[j] != null) {
 138                 out.write(new Format("%-" + maxid + "s").form(s[j].getId()+" "));
 139                 int start = i*len;
 140                 int end = start + len;
 141
 142                 if (end < s[j].getSeq().length() && start < s[j].getSeq().length() ) {
 143                   out.write(s[j].getSeq().substring(start,end) + "\n");
 144                 } else {
 145                   if (start < s[j].getSeq().length()) {
 146                     out.write(s[j].getSeq().substring(start) + "\n");
 147                   }
 148                 }
 149                 j++;
 150               }
 151               out.write("\n");
 152
 153             }
 154          }
 155
 156
 157           public static Alignment make_Alignment(vamsas.objects.simple.Action origin, Sequence[] seqs, String[] source) throws Exception {
 158                 Pattern nonGap = Pattern.compile("[A-Z*0-9]", Pattern.CASE_INSENSITIVE);
 159                 boolean gapsset = false;
 160                 char gapchar='-';
 161                 int seqLength = 0;
 162
 163
 164                 for (int i=0, nseq=seqs.length; i<nseq; i++) {
 165                         String seq = seqs[i].getSeq();
 166                         String gaps = nonGap.matcher(seq).replaceAll("");
 167                         if (seqLength==0) {
 168                                 seqLength=seq.length();
 169                         } else
 170                                 if (seqLength!=seq.length())
 171                                         throw(new Exception(i+"th Sequence (>"+seqs[i].getId()+") is not aligned.\n"));//TODO: move this to assertions part of Alignment
 172
 173                         // common check for any sequence...
 174                         if (gaps!=null && gaps.length()>0) {
 175                                 if (!gapsset)
 176                                         gapchar = gaps.charAt(0);
 177                                 for (int c=0, gc=gaps.length(); c<gc; c++) {
 178                                         if (gapchar!=gaps.charAt(c)) {
 179                                                 throw(new IOException("Inconsistent gap characters in sequence "+i+": '"+seq+"'"));
 180                                         }
 181                                 }
 182                         }
 183                 }
 184
 185                 return new Alignment(origin, new SequenceSet(seqs), source, new String(""+gapchar));
 186           }
 187
 188           public static Alignment read_FastaAlignment(InputStream os, String[] source) throws Exception {
 189                 Sequence[] seqs;
 190                 try {
 191                         seqs = SeqSet.read_SeqFasta(os);
 192                         if (seqs==null)
 193                                 throw(new Exception("Empty alignment stream!\n"));
 194                 } catch (Exception e) {
 195                         throw new Exception("Invalid fasta alignment\n",e);
 196                 }
 197
 198                 return make_Alignment(new vamsas.objects.simple.Action(source[0]), seqs, source);
 199           }
 200           public static Alignment read_ClustalAlignment(InputStream os, String[] source) throws Exception {
 201                 Sequence[] seqs;
 202                 try {
 203                         seqs = SeqAln.ReadClustalFile(os);
 204                         if (seqs==null)
 205                                 throw(new Exception("Empty alignment stream!\n"));
 206                 } catch (Exception e) {
 207                         throw new Exception("Invalid fasta alignment\n",e);
 208                 }
 209
 210                 return make_Alignment(new vamsas.objects.simple.Action(source[0]), seqs, source);
 211           }
 212 }