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
30 public class MSFfile extends AlignFile {
\r
31 private static com.stevesoft.pat.Regex gapre = new com.stevesoft.pat.Regex("\\~",
\r
33 private static com.stevesoft.pat.Regex re2gap = new com.stevesoft.pat.Regex(
\r
34 "[" + jalview.util.Comparison.GapChars + "]", "\\~");
\r
39 public MSFfile(String inStr) {
\r
43 public MSFfile(String inFile, String type) throws IOException {
\r
44 super(inFile, type);
\r
47 public void parse() {
\r
49 boolean seqFlag = false;
\r
50 String key = new String();
\r
51 Vector headers = new Vector();
\r
52 Hashtable seqhash = new Hashtable();
\r
56 while ((line = nextLine()) != null) {
\r
57 StringTokenizer str = new StringTokenizer(line);
\r
59 while (str.hasMoreTokens()) {
\r
60 String inStr = str.nextToken();
\r
62 //If line has header information add to the headers vector
\r
63 if (inStr.indexOf("Name:") != -1) {
\r
64 key = str.nextToken();
\r
65 headers.addElement(key);
\r
68 //if line has // set SeqFlag to 1 so we know sequences are coming
\r
69 if (inStr.indexOf("//") != -1) {
\r
73 //Process lines as sequence lines if seqFlag is set
\r
74 if ((inStr.indexOf("//") == -1) && (seqFlag == true)) {
\r
75 //seqeunce id is the first field
\r
78 StringBuffer tempseq;
\r
80 //Get sequence from hash if it exists
\r
81 if (seqhash.containsKey(key)) {
\r
82 tempseq = (StringBuffer) seqhash.get(key);
\r
84 tempseq = new StringBuffer();
\r
85 seqhash.put(key, tempseq);
\r
88 //loop through the rest of the words
\r
89 while (str.hasMoreTokens()) {
\r
90 //append the word to the sequence
\r
91 tempseq.append(str.nextToken());
\r
96 } catch (IOException e) {
\r
97 System.err.println("Exception parsing MSFFile " + e);
\r
98 e.printStackTrace();
\r
101 this.noSeqs = headers.size();
\r
103 //Add sequences to the hash
\r
104 for (i = 0; i < headers.size(); i++) {
\r
105 if (seqhash.get(headers.elementAt(i)) != null) {
\r
106 String head = headers.elementAt(i).toString();
\r
107 String seq = seqhash.get(head).toString();
\r
110 int end = seq.length();
\r
112 if (maxLength < head.length()) {
\r
113 maxLength = head.length();
\r
116 if (head.indexOf("/") > 0) {
\r
117 StringTokenizer st = new StringTokenizer(head, "/");
\r
119 if (st.countTokens() == 2) {
\r
120 head = st.nextToken();
\r
122 String tmp = st.nextToken();
\r
123 st = new StringTokenizer(tmp, "-");
\r
125 if (st.countTokens() == 2) {
\r
126 start = Integer.valueOf(st.nextToken()).intValue();
\r
127 end = Integer.valueOf(st.nextToken()).intValue();
\r
132 // Replace ~ with a sensible gap character
\r
133 seq = gapre.replaceAll(seq);
\r
135 Sequence newSeq = new Sequence(head, seq, start, end);
\r
137 seqs.addElement(newSeq);
\r
139 System.err.println("MSFFile Parser: Can't find sequence for " +
\r
140 headers.elementAt(i));
\r
145 public static int checkSum(String seq) {
\r
146 //String chars = "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz.*~&@";
\r
149 String index = "--------------------------------------&---*---.-----------------@ABCDEFGHIJKLMNOPQRSTUVWXYZ------ABCDEFGHIJKLMNOPQRSTUVWXYZ----@";
\r
150 index += "--------------------------------------------------------------------------------------------------------------------------------";
\r
152 for (int i = 0; i < seq.length(); i++) {
\r
154 if (i < seq.length()) {
\r
155 int pos = index.indexOf(seq.substring(i, i + 1));
\r
157 if (!index.substring(pos, pos + 1).equals("_")) {
\r
158 check += (((i % 57) + 1) * pos);
\r
161 } catch (Exception e) {
\r
162 System.err.println("Exception during MSF Checksum calculation");
\r
163 e.printStackTrace();
\r
167 return check % 10000;
\r
170 public static String print(SequenceI[] s) {
\r
171 return print(s, false);
\r
174 public static String print(SequenceI[] s, boolean is_NA) {
\r
175 StringBuffer out = new StringBuffer("!!" + (is_NA ? "NA" : "AA") +
\r
176 "_MULTIPLE_ALIGNMENT 1.0\n\n"); // TODO: JBPNote : Jalview doesn't remember NA or AA yet.
\r
183 while ((i < s.length) && (s[i] != null)) {
\r
185 big += (sq = s[i].getSequence());
\r
187 if (sq.length() > max) {
\r
194 Format maxLenpad = new Format("%" + (new String("" + max)).length() +
\r
196 Format maxChkpad = new Format("%" + (new String("1" + max)).length() +
\r
200 long bigcheck = checkSum(big);
\r
202 out.append(" MSF: " + s[0].getSequence().length() + " Type: " +
\r
203 (is_NA ? "N" : "P") + " Check: " + bigcheck + " ..\n\n\n");
\r
205 String[] nameBlock = new String[s.length];
\r
206 String[] idBlock = new String[s.length];
\r
208 while ((i < s.length) && (s[i] != null)) {
\r
209 String seq = s[i].getSequence();
\r
210 String name = s[i].getName() + "/" + s[i].getStart() + "-" +
\r
212 int check = checkSum(s[i].getSequence());
\r
213 nameBlock[i] = new String(" Name: " + name + " ");
\r
214 idBlock[i] = new String("Len: " +
\r
215 maxLenpad.form(s[i].getSequence().length()) + " Check:" +
\r
216 maxChkpad.form(check) + " Weight: 1.00\n");
\r
218 if (name.length() > maxid) {
\r
219 maxid = name.length();
\r
222 if (nameBlock[i].length() > maxNB) {
\r
223 maxNB = nameBlock[i].length();
\r
237 Format nbFormat = new Format("%-" + maxNB + "s");
\r
239 for (i = 0; (i < s.length) && (s[i] != null); i++) {
\r
240 out.append(nbFormat.form(nameBlock[i]) + idBlock[i]);
\r
244 out.append("\n\n//\n\n");
\r
248 int nochunks = (max / len) + 1;
\r
250 if ((max % len) == 0) {
\r
254 for (i = 0; i < nochunks; i++) {
\r
257 while ((j < s.length) && (s[j] != null)) {
\r
258 String name = s[j].getName();
\r
259 out.append(new Format("%-" + maxid + "s").form(name + "/" +
\r
260 s[j].getStart() + "-" + s[j].getEnd()) + " ");
\r
262 for (int k = 0; k < 5; k++) {
\r
263 int start = (i * 50) + (k * 10);
\r
264 int end = start + 10;
\r
266 if ((end < s[j].getSequence().length()) &&
\r
267 (start < s[j].getSequence().length())) {
\r
268 out.append(re2gap.replaceAll(s[j].getSequence()
\r
269 .substring(start, end)));
\r
272 // out.append(" ");
\r
277 if (start < s[j].getSequence().length()) {
\r
278 out.append(re2gap.replaceAll(
\r
279 s[j].getSequence().substring(start)));
\r
295 return out.toString();
\r
298 public String print() {
\r
299 return print(getSeqsAsArray());
\r