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
22 import jalview.datamodel.*;
\r
23 import jalview.util.*;
\r
28 public class MSFfile extends AlignFile {
\r
33 public MSFfile(String inStr) {
\r
37 public MSFfile(String inFile, String type) throws IOException {
\r
41 private static com.stevesoft.pat.Regex gapre = new com.stevesoft.pat.Regex("\\~","-");
\r
42 private static com.stevesoft.pat.Regex re2gap = new com.stevesoft.pat.Regex("["+jalview.util.Comparison.GapChars+"]","\\~");
\r
44 public void parse() {
\r
46 boolean seqFlag = false;
\r
47 String key = new String();
\r
48 Vector headers = new Vector();
\r
49 Hashtable seqhash = new Hashtable();
\r
53 while ((line = nextLine()) != null) {
\r
55 StringTokenizer str = new StringTokenizer(line);
\r
57 while (str.hasMoreTokens()) {
\r
59 String inStr = str.nextToken();
\r
61 //If line has header information add to the headers vector
\r
62 if (inStr.indexOf("Name:") != -1) {
\r
63 key = str.nextToken();
\r
64 headers.addElement(key);
\r
67 //if line has // set SeqFlag to 1 so we know sequences are coming
\r
68 if (inStr.indexOf("//") != -1) {
\r
72 //Process lines as sequence lines if seqFlag is set
\r
73 if (( inStr.indexOf("//") == -1) && (seqFlag == true)) {
\r
74 //seqeunce id is the first field
\r
76 StringBuffer tempseq;
\r
78 //Get sequence from hash if it exists
\r
79 if (seqhash.containsKey(key)) {
\r
80 tempseq = (StringBuffer)seqhash.get(key);
\r
82 tempseq = new StringBuffer();
\r
83 seqhash.put(key,tempseq);
\r
86 //loop through the rest of the words
\r
87 while (str.hasMoreTokens()) {
\r
88 //append the word to the sequence
\r
89 tempseq.append(str.nextToken());
\r
94 } catch (IOException e) {
\r
95 System.err.println("Exception parsing MSFFile " + e);
\r
96 e.printStackTrace();
\r
99 this.noSeqs = headers.size();
\r
101 //Add sequences to the hash
\r
102 for (i = 0; i < headers.size(); i++ ) {
\r
104 if ( seqhash.get(headers.elementAt(i)) != null) {
\r
105 String head = headers.elementAt(i).toString();
\r
106 String seq = seqhash.get(head).toString();
\r
109 int end = seq.length();
\r
111 if (maxLength < head.length() ) {
\r
112 maxLength = head.length();
\r
115 if (head.indexOf("/") > 0 ) {
\r
117 StringTokenizer st = new StringTokenizer(head,"/");
\r
119 if (st.countTokens() == 2) {
\r
121 head = st.nextToken();
\r
122 String tmp = st.nextToken();
\r
123 st = new StringTokenizer(tmp,"-");
\r
124 if (st.countTokens() == 2) {
\r
125 start = Integer.valueOf(st.nextToken()).intValue();
\r
126 end = Integer.valueOf(st.nextToken()).intValue();
\r
130 // Replace ~ with a sensible gap character
\r
131 seq = gapre.replaceAll(seq);
\r
132 Sequence newSeq = new Sequence(head,seq,start,end);
\r
134 seqs.addElement(newSeq);
\r
137 System.err.println("MSFFile Parser: Can't find sequence for " + headers.elementAt(i));
\r
143 public static int checkSum(String seq) {
\r
144 //String chars = "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz.*~&@";
\r
147 String index = "--------------------------------------&---*---.-----------------@ABCDEFGHIJKLMNOPQRSTUVWXYZ------ABCDEFGHIJKLMNOPQRSTUVWXYZ----@";
\r
148 index += "--------------------------------------------------------------------------------------------------------------------------------";
\r
150 for(int i = 0; i < seq.length(); i++) {
\r
152 if (i <seq.length()) {
\r
153 int pos = index.indexOf(seq.substring(i,i+1));
\r
154 if (!index.substring(pos,pos+1).equals("_")) {
\r
155 check += ((i % 57) + 1) * pos;
\r
158 } catch (Exception e) {
\r
159 System.err.println("Exception during MSF Checksum calculation");
\r
160 e.printStackTrace();
\r
163 return check % 10000;
\r
166 public static String print(SequenceI[] s) {
\r
167 return print(s, false);
\r
169 public static String print(SequenceI[] s, boolean is_NA) {
\r
170 StringBuffer out = new StringBuffer("!!"+(is_NA ? "NA":"AA")+"_MULTIPLE_ALIGNMENT 1.0\n\n"); // TODO: JBPNote : Jalview doesn't remember NA or AA yet.
\r
177 while (i < s.length && s[i] != null) {
\r
179 big += (sq=s[i].getSequence());
\r
180 if (sq.length() > max) {
\r
185 Format maxLenpad = new Format("%"+(new String(""+max)).length()+"d");
\r
186 Format maxChkpad = new Format("%"+(new String("1"+max)).length()+"d");
\r
188 long bigcheck = checkSum(big);
\r
190 out.append(" MSF: " + s[0].getSequence().length() + " Type: "+(is_NA?"N":"P")+" Check: " + bigcheck + " ..\n\n\n");
\r
191 String nameBlock[] = new String[s.length];
\r
192 String idBlock[] = new String[s.length];
\r
193 while (i < s.length && s[i] != null) {
\r
194 String seq = s[i].getSequence();
\r
195 String name = s[i].getName()+ "/" + s[i].getStart() + "-" + s[i].getEnd();
\r
196 int check = checkSum(s[i].getSequence());
\r
197 nameBlock[i]=new String(" Name: "+name+" ");
\r
198 idBlock[i] = new String("Len: " + maxLenpad.form(s[i].getSequence().length()) + " Check:" + maxChkpad.form(check) + " Weight: 1.00\n");
\r
200 if (name.length() > maxid) {
\r
201 maxid = name.length();
\r
203 if (nameBlock[i].length()>maxNB) {
\r
204 maxNB=nameBlock[i].length();
\r
215 Format nbFormat = new Format("%-"+maxNB+"s");
\r
216 for (i=0;i<s.length && s[i]!=null;i++) {
\r
217 out.append(nbFormat.form(nameBlock[i])+idBlock[i]);
\r
220 out.append( "\n\n//\n\n");
\r
224 int nochunks = max / len + 1;
\r
225 if (max%len == 0) {
\r
228 for (i = 0; i < nochunks; i++) {
\r
230 while (j < s.length && s[j] != null) {
\r
231 String name = s[j].getName();
\r
232 out.append( new Format("%-" + maxid + "s").form(name + "/" + s[j].getStart() + "-" + s[j].getEnd()) + " ");
\r
233 for (int k = 0; k < 5; k++) {
\r
235 int start = i*50 + k*10;
\r
236 int end = start + 10;
\r
238 if (end < s[j].getSequence().length() && start < s[j].getSequence().length() ) {
\r
239 out.append(re2gap.replaceAll(s[j].getSequence().substring(start,end)));
\r
241 // out.append(" ");
\r
246 if (start < s[j].getSequence().length()) {
\r
247 out.append(re2gap.replaceAll(s[j].getSequence().substring(start)));
\r
261 return out.toString();
\r
263 public String print() {
\r
264 return print(getSeqsAsArray());
\r