2 * Jalview - A Sequence Alignment Editor and Viewer
\r
3 * Copyright (C) 2005 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
21 import jalview.datamodel.*;
\r
23 import jalview.util.*;
\r
34 * @version $Revision$
\r
36 public class MSFfile extends AlignFile
\r
38 private static com.stevesoft.pat.Regex gapre = new com.stevesoft.pat.Regex("\\~",
\r
40 private static com.stevesoft.pat.Regex re2gap = new com.stevesoft.pat.Regex(
\r
41 "[" + jalview.util.Comparison.GapChars + "]", "\\~");
\r
44 * Creates a new MSFfile object.
\r
51 * Creates a new MSFfile object.
\r
53 * @param inStr DOCUMENT ME!
\r
55 public MSFfile(String inStr)
\r
61 * Creates a new MSFfile object.
\r
63 * @param inFile DOCUMENT ME!
\r
64 * @param type DOCUMENT ME!
\r
66 * @throws IOException DOCUMENT ME!
\r
68 public MSFfile(String inFile, String type) throws IOException
\r
70 super(inFile, type);
\r
79 boolean seqFlag = false;
\r
80 String key = new String();
\r
81 Vector headers = new Vector();
\r
82 Hashtable seqhash = new Hashtable();
\r
87 while ((line = nextLine()) != null)
\r
89 StringTokenizer str = new StringTokenizer(line);
\r
91 while (str.hasMoreTokens())
\r
93 String inStr = str.nextToken();
\r
95 //If line has header information add to the headers vector
\r
96 if (inStr.indexOf("Name:") != -1)
\r
98 key = str.nextToken();
\r
99 headers.addElement(key);
\r
102 //if line has // set SeqFlag to 1 so we know sequences are coming
\r
103 if (inStr.indexOf("//") != -1)
\r
108 //Process lines as sequence lines if seqFlag is set
\r
109 if ((inStr.indexOf("//") == -1) && (seqFlag == true))
\r
111 //seqeunce id is the first field
\r
114 StringBuffer tempseq;
\r
116 //Get sequence from hash if it exists
\r
117 if (seqhash.containsKey(key))
\r
119 tempseq = (StringBuffer) seqhash.get(key);
\r
123 tempseq = new StringBuffer();
\r
124 seqhash.put(key, tempseq);
\r
127 //loop through the rest of the words
\r
128 while (str.hasMoreTokens())
\r
130 //append the word to the sequence
\r
131 tempseq.append(str.nextToken());
\r
137 catch (IOException e)
\r
139 System.err.println("Exception parsing MSFFile " + e);
\r
140 e.printStackTrace();
\r
143 this.noSeqs = headers.size();
\r
145 //Add sequences to the hash
\r
146 for (i = 0; i < headers.size(); i++)
\r
148 if (seqhash.get(headers.elementAt(i)) != null)
\r
150 String head = headers.elementAt(i).toString();
\r
151 String seq = seqhash.get(head).toString();
\r
156 if (maxLength < head.length())
\r
158 maxLength = head.length();
\r
161 if (head.indexOf("/") > 0)
\r
163 StringTokenizer st = new StringTokenizer(head, "/");
\r
165 if (st.countTokens() == 2)
\r
167 head = st.nextToken();
\r
169 String tmp = st.nextToken();
\r
170 st = new StringTokenizer(tmp, "-");
\r
172 if (st.countTokens() == 2)
\r
174 start = Integer.valueOf(st.nextToken()).intValue();
\r
175 end = Integer.valueOf(st.nextToken()).intValue();
\r
180 // Replace ~ with a sensible gap character
\r
181 seq = gapre.replaceAll(seq);
\r
183 Sequence newSeq = new Sequence(head, seq, start, end);
\r
185 seqs.addElement(newSeq);
\r
189 System.err.println("MSFFile Parser: Can't find sequence for " +
\r
190 headers.elementAt(i));
\r
198 * @param seq DOCUMENT ME!
\r
200 * @return DOCUMENT ME!
\r
202 public static int checkSum(String seq)
\r
204 //String chars = "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz.*~&@";
\r
206 String sequence = seq.toUpperCase();
\r
208 String index = "--------------------------------------&---*---.-----------------@ABCDEFGHIJKLMNOPQRSTUVWXYZ------ABCDEFGHIJKLMNOPQRSTUVWXYZ----@";
\r
209 index += "--------------------------------------------------------------------------------------------------------------------------------";
\r
211 for (int i = 0; i < sequence.length(); i++)
\r
215 int pos = index.indexOf(sequence.charAt(i));
\r
217 if (index.charAt(pos)!='_')
\r
219 check += (((i % 57) + 1) * pos);
\r
222 catch (Exception e)
\r
224 System.err.println("Exception during MSF Checksum calculation");
\r
225 e.printStackTrace();
\r
229 return check % 10000;
\r
235 * @param s DOCUMENT ME!
\r
237 * @return DOCUMENT ME!
\r
239 public static String print(SequenceI[] s)
\r
241 return print(s, false);
\r
247 * @param s DOCUMENT ME!
\r
248 * @param is_NA DOCUMENT ME!
\r
250 * @return DOCUMENT ME!
\r
252 public static String print(SequenceI[] s, boolean is_NA)
\r
254 StringBuffer out = new StringBuffer("!!" + (is_NA ? "NA" : "AA") +
\r
255 "_MULTIPLE_ALIGNMENT 1.0\n\n"); // TODO: JBPNote : Jalview doesn't remember NA or AA yet.
\r
262 while ((i < s.length) && (s[i] != null))
\r
265 big += (sq = s[i].getSequence());
\r
267 if (sq.length() > max)
\r
275 Format maxLenpad = new Format("%" + (new String("" + max)).length() +
\r
277 Format maxChkpad = new Format("%" + (new String("1" + max)).length() +
\r
281 long bigcheck = checkSum(big);
\r
283 out.append(" MSF: " + s[0].getSequence().length() + " Type: " +
\r
284 (is_NA ? "N" : "P") + " Check: " + bigcheck + " ..\n\n\n");
\r
286 String[] nameBlock = new String[s.length];
\r
287 String[] idBlock = new String[s.length];
\r
289 while ((i < s.length) && (s[i] != null))
\r
291 String name = s[i].getName() + "/" + s[i].getStart() + "-" +
\r
293 int check = checkSum(s[i].getSequence());
\r
294 nameBlock[i] = new String(" Name: " + name + " ");
\r
295 idBlock[i] = new String("Len: " +
\r
296 maxLenpad.form(s[i].getSequence().length()) + " Check:" +
\r
297 maxChkpad.form(check) + " Weight: 1.00\n");
\r
299 if (name.length() > maxid)
\r
301 maxid = name.length();
\r
304 if (nameBlock[i].length() > maxNB)
\r
306 maxNB = nameBlock[i].length();
\r
322 Format nbFormat = new Format("%-" + maxNB + "s");
\r
324 for (i = 0; (i < s.length) && (s[i] != null); i++)
\r
326 out.append(nbFormat.form(nameBlock[i]) + idBlock[i]);
\r
330 out.append("\n\n//\n\n");
\r
334 int nochunks = (max / len) + 1;
\r
336 if ((max % len) == 0)
\r
341 for (i = 0; i < nochunks; i++)
\r
345 while ((j < s.length) && (s[j] != null))
\r
347 String name = s[j].getName();
\r
348 out.append(new Format("%-" + maxid + "s").form(name + "/" +
\r
349 s[j].getStart() + "-" + s[j].getEnd()) + " ");
\r
351 for (int k = 0; k < 5; k++)
\r
353 int start = (i * 50) + (k * 10);
\r
354 int end = start + 10;
\r
356 if ((end < s[j].getSequence().length()) &&
\r
357 (start < s[j].getSequence().length()))
\r
359 out.append(re2gap.replaceAll(s[j].getSequence()
\r
360 .substring(start, end)));
\r
364 // out.append(" ");
\r
373 if (start < s[j].getSequence().length())
\r
375 out.append(re2gap.replaceAll(
\r
376 s[j].getSequence().substring(start)));
\r
395 return out.toString();
\r
401 * @return DOCUMENT ME!
\r
403 public String print()
\r
405 return print(getSeqsAsArray());
\r