3 import jalview.datamodel.*;
\r
4 import jalview.util.*;
\r
9 public class MSFfile extends AlignFile {
\r
14 public MSFfile(String inStr) {
\r
18 public MSFfile(String inFile, String type) throws IOException {
\r
22 private static com.stevesoft.pat.Regex gapre = new com.stevesoft.pat.Regex("\\~","-");
\r
23 private static com.stevesoft.pat.Regex re2gap = new com.stevesoft.pat.Regex("["+jalview.util.Comparison.GapChars+"]","\\~");
\r
25 public void parse() {
\r
27 boolean seqFlag = false;
\r
28 String key = new String();
\r
29 Vector headers = new Vector();
\r
30 Hashtable seqhash = new Hashtable();
\r
34 while ((line = nextLine()) != null) {
\r
36 StringTokenizer str = new StringTokenizer(line);
\r
38 while (str.hasMoreTokens()) {
\r
40 String inStr = str.nextToken();
\r
42 //If line has header information add to the headers vector
\r
43 if (inStr.indexOf("Name:") != -1) {
\r
44 key = str.nextToken();
\r
45 headers.addElement(key);
\r
48 //if line has // set SeqFlag to 1 so we know sequences are coming
\r
49 if (inStr.indexOf("//") != -1) {
\r
53 //Process lines as sequence lines if seqFlag is set
\r
54 if (( inStr.indexOf("//") == -1) && (seqFlag == true)) {
\r
55 //seqeunce id is the first field
\r
57 StringBuffer tempseq;
\r
59 //Get sequence from hash if it exists
\r
60 if (seqhash.containsKey(key)) {
\r
61 tempseq = (StringBuffer)seqhash.get(key);
\r
63 tempseq = new StringBuffer();
\r
64 seqhash.put(key,tempseq);
\r
67 //loop through the rest of the words
\r
68 while (str.hasMoreTokens()) {
\r
69 //append the word to the sequence
\r
70 tempseq.append(str.nextToken());
\r
75 } catch (IOException e) {
\r
76 System.err.println("Exception parsing MSFFile " + e);
\r
77 e.printStackTrace();
\r
80 this.noSeqs = headers.size();
\r
82 //Add sequences to the hash
\r
83 for (i = 0; i < headers.size(); i++ ) {
\r
85 if ( seqhash.get(headers.elementAt(i)) != null) {
\r
86 String head = headers.elementAt(i).toString();
\r
87 String seq = seqhash.get(head).toString();
\r
90 int end = seq.length();
\r
92 if (maxLength < head.length() ) {
\r
93 maxLength = head.length();
\r
96 if (head.indexOf("/") > 0 ) {
\r
98 StringTokenizer st = new StringTokenizer(head,"/");
\r
100 if (st.countTokens() == 2) {
\r
102 head = st.nextToken();
\r
103 String tmp = st.nextToken();
\r
104 st = new StringTokenizer(tmp,"-");
\r
105 if (st.countTokens() == 2) {
\r
106 start = Integer.valueOf(st.nextToken()).intValue();
\r
107 end = Integer.valueOf(st.nextToken()).intValue();
\r
111 // Replace ~ with a sensible gap character
\r
112 seq = gapre.replaceAll(seq);
\r
113 Sequence newSeq = new Sequence(head,seq,start,end);
\r
115 seqs.addElement(newSeq);
\r
118 System.err.println("MSFFile Parser: Can't find sequence for " + headers.elementAt(i));
\r
124 public static int checkSum(String seq) {
\r
125 //String chars = "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz.*~&@";
\r
128 String index = "--------------------------------------&---*---.-----------------@ABCDEFGHIJKLMNOPQRSTUVWXYZ------ABCDEFGHIJKLMNOPQRSTUVWXYZ----@";
\r
129 index += "--------------------------------------------------------------------------------------------------------------------------------";
\r
131 for(int i = 0; i < seq.length(); i++) {
\r
133 if (i <seq.length()) {
\r
134 int pos = index.indexOf(seq.substring(i,i+1));
\r
135 if (!index.substring(pos,pos+1).equals("_")) {
\r
136 check += ((i % 57) + 1) * pos;
\r
139 } catch (Exception e) {
\r
140 System.err.println("Exception during MSF Checksum calculation");
\r
141 e.printStackTrace();
\r
144 return check % 10000;
\r
147 public static String print(SequenceI[] s) {
\r
148 return print(s, false);
\r
150 public static String print(SequenceI[] s, boolean is_NA) {
\r
151 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
158 while (i < s.length && s[i] != null) {
\r
160 big += (sq=s[i].getSequence());
\r
161 if (sq.length() > max) {
\r
166 Format maxLenpad = new Format("%"+(new String(""+max)).length()+"d");
\r
167 Format maxChkpad = new Format("%"+(new String("1"+max)).length()+"d");
\r
169 long bigcheck = checkSum(big);
\r
171 out.append(" MSF: " + s[0].getSequence().length() + " Type: "+(is_NA?"N":"P")+" Check: " + bigcheck + " ..\n\n\n");
\r
172 String nameBlock[] = new String[s.length];
\r
173 String idBlock[] = new String[s.length];
\r
174 while (i < s.length && s[i] != null) {
\r
175 String seq = s[i].getSequence();
\r
176 String name = s[i].getName()+ "/" + s[i].getStart() + "-" + s[i].getEnd();
\r
177 int check = checkSum(s[i].getSequence());
\r
178 nameBlock[i]=new String(" Name: "+name+" ");
\r
179 idBlock[i] = new String("Len: " + maxLenpad.form(s[i].getSequence().length()) + " Check:" + maxChkpad.form(check) + " Weight: 1.00\n");
\r
181 if (name.length() > maxid) {
\r
182 maxid = name.length();
\r
184 if (nameBlock[i].length()>maxNB) {
\r
185 maxNB=nameBlock[i].length();
\r
196 Format nbFormat = new Format("%-"+maxNB+"s");
\r
197 for (i=0;i<s.length && s[i]!=null;i++) {
\r
198 out.append(nbFormat.form(nameBlock[i])+idBlock[i]);
\r
201 out.append( "\n\n//\n\n");
\r
205 int nochunks = max / len + 1;
\r
206 if (max%len == 0) {
\r
209 for (i = 0; i < nochunks; i++) {
\r
211 while (j < s.length && s[j] != null) {
\r
212 String name = s[j].getName();
\r
213 out.append( new Format("%-" + maxid + "s").form(name + "/" + s[j].getStart() + "-" + s[j].getEnd()) + " ");
\r
214 for (int k = 0; k < 5; k++) {
\r
216 int start = i*50 + k*10;
\r
217 int end = start + 10;
\r
219 if (end < s[j].getSequence().length() && start < s[j].getSequence().length() ) {
\r
220 out.append(re2gap.replaceAll(s[j].getSequence().substring(start,end)));
\r
222 // out.append(" ");
\r
227 if (start < s[j].getSequence().length()) {
\r
228 out.append(re2gap.replaceAll(s[j].getSequence().substring(start)));
\r
242 return out.toString();
\r
244 public String print() {
\r
245 return print(getSeqsAsArray());
\r