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 public void parse() {
\r
25 boolean seqFlag = false;
\r
26 String key = new String();
\r
27 Vector headers = new Vector();
\r
28 Hashtable seqhash = new Hashtable();
\r
32 while ((line = nextLine()) != null) {
\r
34 StringTokenizer str = new StringTokenizer(line);
\r
36 while (str.hasMoreTokens()) {
\r
38 String inStr = str.nextToken();
\r
40 //If line has header information add to the headers vector
\r
41 if (inStr.indexOf("Name:") != -1) {
\r
42 key = str.nextToken();
\r
43 headers.addElement(key);
\r
46 //if line has // set SeqFlag to 1 so we know sequences are coming
\r
47 if (inStr.indexOf("//") != -1) {
\r
51 //Process lines as sequence lines if seqFlag is set
\r
52 if (( inStr.indexOf("//") == -1) && (seqFlag == true)) {
\r
53 //seqeunce id is the first field
\r
55 StringBuffer tempseq;
\r
57 //Get sequence from hash if it exists
\r
58 if (seqhash.containsKey(key)) {
\r
59 tempseq = (StringBuffer)seqhash.get(key);
\r
61 tempseq = new StringBuffer();
\r
62 seqhash.put(key,tempseq);
\r
65 //loop through the rest of the words
\r
66 while (str.hasMoreTokens()) {
\r
67 //append the word to the sequence
\r
68 tempseq.append(str.nextToken());
\r
73 } catch (IOException e) {
\r
74 System.out.println("Exception parsing MSFFile " + e);
\r
77 this.noSeqs = headers.size();
\r
79 //Add sequences to the hash
\r
80 for (i = 0; i < headers.size(); i++ ) {
\r
82 if ( seqhash.get(headers.elementAt(i)) != null) {
\r
83 String head = headers.elementAt(i).toString();
\r
84 String seq = seqhash.get(head).toString();
\r
87 int end = seq.length();
\r
89 if (maxLength < head.length() ) {
\r
90 maxLength = head.length();
\r
93 if (head.indexOf("/") > 0 ) {
\r
95 StringTokenizer st = new StringTokenizer(head,"/");
\r
97 if (st.countTokens() == 2) {
\r
99 head = st.nextToken();
\r
100 String tmp = st.nextToken();
\r
101 st = new StringTokenizer(tmp,"-");
\r
102 if (st.countTokens() == 2) {
\r
103 start = Integer.valueOf(st.nextToken()).intValue();
\r
104 end = Integer.valueOf(st.nextToken()).intValue();
\r
109 Sequence newSeq = new Sequence(head,seq,start,end);
\r
111 seqs.addElement(newSeq);
\r
114 System.out.println("Can't find sequence for " + headers.elementAt(i));
\r
120 public static int checkSum(String seq) {
\r
121 //String chars = "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz.*~&@";
\r
124 String index = "--------------------------------------&---*---.-----------------@ABCDEFGHIJKLMNOPQRSTUVWXYZ------ABCDEFGHIJKLMNOPQRSTUVWXYZ----@";
\r
125 index += "--------------------------------------------------------------------------------------------------------------------------------";
\r
127 for(int i = 0; i < seq.length(); i++) {
\r
129 if (i <seq.length()) {
\r
130 int pos = index.indexOf(seq.substring(i,i+1));
\r
131 if (!index.substring(pos,pos+1).equals("_")) {
\r
132 check += ((i % 57) + 1) * pos;
\r
135 } catch (Exception e) {
\r
136 System.err.println("Exception " + e);
\r
139 return check % 10000;
\r
142 public static String print(SequenceI[] s) {
\r
143 StringBuffer out = new StringBuffer("PileUp\n\n");
\r
150 while (i < s.length && s[i] != null) {
\r
151 big += s[i].getSequence();
\r
155 int bigcheck = checkSum(big);
\r
157 out.append(" MSF: " + s[0].getSequence().length() + " Type: P Check: " + bigcheck + " ..\n\n\n");
\r
159 while (i < s.length && s[i] != null) {
\r
160 String seq = s[i].getSequence();
\r
161 String name = s[i].getName()+ "/" + s[i].getStart() + "-" + s[i].getEnd();
\r
162 int check = checkSum(s[i].getSequence());
\r
163 out.append(" Name: " + name + " oo Len: " + s[i].getSequence().length() + " Check: " + check + " Weight: 1.00\n");
\r
164 if (seq.length() > max) {
\r
165 max = seq.length();
\r
167 if (name.length() > maxid) {
\r
168 maxid = name.length();
\r
177 out.append( "\n\n//\n\n");
\r
181 int nochunks = max / len + 1;
\r
182 if (max%len == 0) {
\r
185 for (i = 0; i < nochunks; i++) {
\r
187 while (j < s.length && s[j] != null) {
\r
188 String name = s[j].getName();
\r
189 out.append( new Format("%-" + maxid + "s").form(name + "/" + s[j].getStart() + "-" + s[j].getEnd()) + " ");
\r
190 for (int k = 0; k < 5; k++) {
\r
192 int start = i*50 + k*10;
\r
193 int end = start + 10;
\r
195 if (end < s[j].getSequence().length() && start < s[j].getSequence().length() ) {
\r
196 out.append(s[j].getSequence().substring(start,end));
\r
203 if (start < s[j].getSequence().length()) {
\r
204 out.append(s[j].getSequence().substring(start));
\r
218 return out.toString();
\r
220 public String print() {
\r
221 return print(getSeqsAsArray());
\r