2 * Jalview - A Sequence Alignment Editor and Viewer
\r
3 * Copyright (C) 2007 AM Waterhouse, J Procter, G Barton, M Clamp, S Searle
\r
5 * This program is free software; you can redistribute it and/or
\r
6 * modify it under the terms of the GNU General Public License
\r
7 * as published by the Free Software Foundation; either version 2
\r
8 * of the License, or (at your option) any later version.
\r
10 * This program is distributed in the hope that it will be useful,
\r
11 * but WITHOUT ANY WARRANTY; without even the implied warranty of
\r
12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
\r
13 * GNU General Public License for more details.
\r
15 * You should have received a copy of the GNU General Public License
\r
16 * along with this program; if not, write to the Free Software
\r
17 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA
\r
24 import jalview.datamodel.*;
\r
25 import jalview.util.*;
\r
31 * @version $Revision$
\r
33 public class MSFfile
\r
38 * Creates a new MSFfile object.
\r
45 * Creates a new MSFfile object.
\r
47 * @param inFile DOCUMENT ME!
\r
48 * @param type DOCUMENT ME!
\r
50 * @throws IOException DOCUMENT ME!
\r
52 public MSFfile(String inFile, String type)
\r
55 super(inFile, type);
\r
65 boolean seqFlag = false;
\r
66 String key = new String();
\r
67 Vector headers = new Vector();
\r
68 Hashtable seqhash = new Hashtable();
\r
73 while ( (line = nextLine()) != null)
\r
75 StringTokenizer str = new StringTokenizer(line);
\r
77 while (str.hasMoreTokens())
\r
79 String inStr = str.nextToken();
\r
81 //If line has header information add to the headers vector
\r
82 if (inStr.indexOf("Name:") != -1)
\r
84 key = str.nextToken();
\r
85 headers.addElement(key);
\r
88 //if line has // set SeqFlag to 1 so we know sequences are coming
\r
89 if (inStr.indexOf("//") != -1)
\r
94 //Process lines as sequence lines if seqFlag is set
\r
95 if ( (inStr.indexOf("//") == -1) && (seqFlag == true))
\r
97 //seqeunce id is the first field
\r
100 StringBuffer tempseq;
\r
102 //Get sequence from hash if it exists
\r
103 if (seqhash.containsKey(key))
\r
105 tempseq = (StringBuffer) seqhash.get(key);
\r
109 tempseq = new StringBuffer();
\r
110 seqhash.put(key, tempseq);
\r
113 //loop through the rest of the words
\r
114 while (str.hasMoreTokens())
\r
116 //append the word to the sequence
\r
117 tempseq.append(str.nextToken());
\r
123 catch (IOException e)
\r
125 System.err.println("Exception parsing MSFFile " + e);
\r
126 e.printStackTrace();
\r
129 this.noSeqs = headers.size();
\r
131 //Add sequences to the hash
\r
132 for (i = 0; i < headers.size(); i++)
\r
134 if (seqhash.get(headers.elementAt(i)) != null)
\r
136 String head = headers.elementAt(i).toString();
\r
137 String seq = seqhash.get(head).toString();
\r
139 if (maxLength < head.length())
\r
141 maxLength = head.length();
\r
144 // Replace ~ with a sensible gap character
\r
145 seq = seq.replace('~', '-');
\r
147 Sequence newSeq = parseId(head);
\r
149 newSeq.setSequence(seq);
\r
151 seqs.addElement(newSeq);
\r
155 System.err.println("MSFFile Parser: Can't find sequence for " +
\r
156 headers.elementAt(i));
\r
164 * @param seq DOCUMENT ME!
\r
166 * @return DOCUMENT ME!
\r
168 public int checkSum(String seq)
\r
171 String sequence = seq.toUpperCase();
\r
173 for (int i = 0; i < sequence.length(); i++)
\r
178 int value = sequence.charAt(i);
\r
181 check += (i % 57 + 1) * value;
\r
184 catch (Exception e)
\r
186 System.err.println("Exception during MSF Checksum calculation");
\r
187 e.printStackTrace();
\r
191 return check % 10000;
\r
197 * @param s DOCUMENT ME!
\r
198 * @param is_NA DOCUMENT ME!
\r
200 * @return DOCUMENT ME!
\r
202 public String print(SequenceI[] seqs)
\r
205 boolean is_NA = jalview.util.Comparison.isNucleotide(seqs);
\r
207 SequenceI[] s = new SequenceI[seqs.length];
\r
209 StringBuffer out = new StringBuffer("!!" + (is_NA ? "NA" : "AA") +
\r
210 "_MULTIPLE_ALIGNMENT 1.0\n\n"); // TODO: JBPNote : Jalview doesn't remember NA or AA yet.
\r
216 while ( (i < seqs.length) && (seqs[i] != null))
\r
218 // Replace all internal gaps with . and external spaces with ~
\r
219 s[i] = new Sequence(seqs[i].getName(),
\r
220 seqs[i].getSequenceAsString().replace('-', '.'));
\r
222 StringBuffer sb = new StringBuffer();
\r
223 sb.append(s[i].getSequence());
\r
225 for (int ii = 0; ii < sb.length(); ii++)
\r
227 if (sb.charAt(ii) == '.')
\r
229 sb.setCharAt(ii, '~');
\r
237 for (int ii = sb.length() - 1; ii > 0; ii--)
\r
239 if (sb.charAt(ii) == '.')
\r
241 sb.setCharAt(ii, '~');
\r
249 s[i].setSequence(sb.toString());
\r
251 if (s[i].getSequence().length > max)
\r
253 max = s[i].getSequence().length;
\r
259 Format maxLenpad = new Format("%" + (new String("" + max)).length() +
\r
261 Format maxChkpad = new Format("%" + (new String("1" + max)).length() +
\r
265 int bigChecksum = 0;
\r
266 int[] checksums = new int[s.length];
\r
267 while (i < s.length)
\r
269 checksums[i] = checkSum(s[i].getSequenceAsString());
\r
270 bigChecksum += checksums[i];
\r
275 out.append(" MSF: " + s[0].getSequence().length + " Type: " +
\r
276 (is_NA ? "N" : "P") + " Check: " + (bigChecksum % 10000) +
\r
279 String[] nameBlock = new String[s.length];
\r
280 String[] idBlock = new String[s.length];
\r
283 while ( (i < s.length) && (s[i] != null))
\r
286 nameBlock[i] = new String(" Name: " + printId(s[i]) + " ");
\r
288 idBlock[i] = new String("Len: " +
\r
289 maxLenpad.form(s[i].getSequence().length) +
\r
291 maxChkpad.form(checksums[i]) + " Weight: 1.00\n");
\r
293 if (s[i].getName().length() > maxid)
\r
295 maxid = s[i].getName().length();
\r
298 if (nameBlock[i].length() > maxNB)
\r
300 maxNB = nameBlock[i].length();
\r
316 Format nbFormat = new Format("%-" + maxNB + "s");
\r
318 for (i = 0; (i < s.length) && (s[i] != null); i++)
\r
320 out.append(nbFormat.form(nameBlock[i]) + idBlock[i]);
\r
324 out.append("\n\n//\n\n");
\r
328 int nochunks = (max / len) + 1;
\r
330 if ( (max % len) == 0)
\r
335 for (i = 0; i < nochunks; i++)
\r
339 while ( (j < s.length) && (s[j] != null))
\r
341 String name = printId(s[j]);
\r
343 out.append(new Format("%-" + maxid + "s").form(name + " "));
\r
345 for (int k = 0; k < 5; k++)
\r
347 int start = (i * 50) + (k * 10);
\r
348 int end = start + 10;
\r
350 if ( (end < s[j].getSequence().length) &&
\r
351 (start < s[j].getSequence().length))
\r
353 out.append(s[j].getSequence(start, end));
\r
366 if (start < s[j].getSequence().length)
\r
368 out.append(s[j].getSequenceAsString().substring(start));
\r
387 return out.toString();
\r
393 * @return DOCUMENT ME!
\r
395 public String print()
\r
397 return print(getSeqsAsArray());
\r