Title:
- * PileUpfile - *Description:
- * - * Read and write PileUp style MSF Files. - * This used to be the MSFFile class, and was written according to the EBI's idea - * of a subset of the MSF alignment format. But, that was updated to reflect current - * GCG style IO fashion, as found in Emboss (thanks David Martin!) - * +/** + *Title:
+ * PileUpfile + *Description:
+ * + * Read and write PileUp style MSF Files. + * This used to be the MSFFile class, and was written according to the EBI's idea + * of a subset of the MSF alignment format. But, that was updated to reflect current + * GCG style IO fashion, as found in Emboss (thanks David Martin!) + * **/ -import jalview.datamodel.*; - -import jalview.util.*; - import java.io.*; - import java.util.*; +import jalview.datamodel.*; +import jalview.util.*; -public class PileUpfile extends AlignFile { - public PileUpfile() { - } - - public PileUpfile(String inStr) { - super(inStr); - } - - public PileUpfile(String inFile, String type) throws IOException { - super(inFile, type); - } +public class PileUpfile + extends AlignFile +{ + public PileUpfile() + { + } + + public PileUpfile(String inStr) + { + super(inStr); + } + + public PileUpfile(String inFile, String type) + throws IOException + { + super(inFile, type); + } + + public void parse() + { + int i = 0; + boolean seqFlag = false; + String key = new String(); + Vector headers = new Vector(); + Hashtable seqhash = new Hashtable(); + String line; + + try + { + while ( (line = nextLine()) != null) + { + StringTokenizer str = new StringTokenizer(line); + + while (str.hasMoreTokens()) + { + String inStr = str.nextToken(); + + //If line has header information add to the headers vector + if (inStr.indexOf("Name:") != -1) + { + key = str.nextToken(); + headers.addElement(key); + } + + //if line has // set SeqFlag to 1 so we know sequences are coming + if (inStr.indexOf("//") != -1) + { + seqFlag = true; + } + + //Process lines as sequence lines if seqFlag is set + if ( (inStr.indexOf("//") == -1) && (seqFlag == true)) + { + //seqeunce id is the first field + key = inStr; + + StringBuffer tempseq; + + //Get sequence from hash if it exists + if (seqhash.containsKey(key)) + { + tempseq = (StringBuffer) seqhash.get(key); + } + else + { + tempseq = new StringBuffer(); + seqhash.put(key, tempseq); + } - public void parse() { - int i = 0; - boolean seqFlag = false; - String key = new String(); - Vector headers = new Vector(); - Hashtable seqhash = new Hashtable(); - String line; - - try { - while ((line = nextLine()) != null) { - StringTokenizer str = new StringTokenizer(line); - - while (str.hasMoreTokens()) { - String inStr = str.nextToken(); - - //If line has header information add to the headers vector - if (inStr.indexOf("Name:") != -1) { - key = str.nextToken(); - headers.addElement(key); - } - - //if line has // set SeqFlag to 1 so we know sequences are coming - if (inStr.indexOf("//") != -1) { - seqFlag = true; - } - - //Process lines as sequence lines if seqFlag is set - if ((inStr.indexOf("//") == -1) && (seqFlag == true)) { - //seqeunce id is the first field - key = inStr; - - StringBuffer tempseq; - - //Get sequence from hash if it exists - if (seqhash.containsKey(key)) { - tempseq = (StringBuffer) seqhash.get(key); - } else { - tempseq = new StringBuffer(); - seqhash.put(key, tempseq); - } - - //loop through the rest of the words - while (str.hasMoreTokens()) { - //append the word to the sequence - tempseq.append(str.nextToken()); - } - } - } + //loop through the rest of the words + while (str.hasMoreTokens()) + { + //append the word to the sequence + tempseq.append(str.nextToken()); } - } catch (IOException e) { - System.err.println("Exception parsing PileUpfile " + e); - e.printStackTrace(); + } } + } + } + catch (IOException e) + { + System.err.println("Exception parsing PileUpfile " + e); + e.printStackTrace(); + } - this.noSeqs = headers.size(); - - //Add sequences to the hash - for (i = 0; i < headers.size(); i++) { - if (seqhash.get(headers.elementAt(i)) != null) { - String head = headers.elementAt(i).toString(); - String seq = seqhash.get(head).toString(); - - int start = 1; - int end = seq.length(); + this.noSeqs = headers.size(); - if (maxLength < head.length()) { - maxLength = head.length(); - } + //Add sequences to the hash + for (i = 0; i < headers.size(); i++) + { + if (seqhash.get(headers.elementAt(i)) != null) + { + String head = headers.elementAt(i).toString(); + String seq = seqhash.get(head).toString(); - if (head.indexOf("/") > 0) { - StringTokenizer st = new StringTokenizer(head, "/"); + int start = 1; + int end = -1; - if (st.countTokens() == 2) { - head = st.nextToken(); + if (maxLength < head.length()) + { + maxLength = head.length(); + } - String tmp = st.nextToken(); - st = new StringTokenizer(tmp, "-"); + if (head.indexOf("/") > 0) + { + StringTokenizer st = new StringTokenizer(head, "/"); - if (st.countTokens() == 2) { - start = Integer.valueOf(st.nextToken()).intValue(); - end = Integer.valueOf(st.nextToken()).intValue(); - } - } - } + if (st.countTokens() == 2) + { + head = st.nextToken(); - Sequence newSeq = new Sequence(head, seq, start, end); + String tmp = st.nextToken(); + st = new StringTokenizer(tmp, "-"); - seqs.addElement(newSeq); - } else { - System.err.println( - "PileUpfile Parser: Can't find sequence for " + - headers.elementAt(i)); + if (st.countTokens() == 2) + { + start = Integer.valueOf(st.nextToken()).intValue(); + end = Integer.valueOf(st.nextToken()).intValue(); } + } } - } - public static int checkSum(String seq) { - //String chars = "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz.*~&@"; - int check = 0; - - String index = "--------------------------------------&---*---.-----------------@ABCDEFGHIJKLMNOPQRSTUVWXYZ------ABCDEFGHIJKLMNOPQRSTUVWXYZ----@"; - index += "--------------------------------------------------------------------------------------------------------------------------------"; - - for (int i = 0; i < seq.length(); i++) { - try { - if (i < seq.length()) { - int pos = index.indexOf(seq.substring(i, i + 1)); - - if (!index.substring(pos, pos + 1).equals("_")) { - check += (((i % 57) + 1) * pos); - } - } - } catch (Exception e) { - System.err.println("Exception during MSF Checksum calculation"); - e.printStackTrace(); - } - } + Sequence newSeq = new Sequence(head, seq, start, end); - return check % 10000; + seqs.addElement(newSeq); + } + else + { + System.err.println( + "PileUpfile Parser: Can't find sequence for " + + headers.elementAt(i)); + } + } + } + + public static int checkSum(String seq) + { + //String chars = "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz.*~&@"; + int check = 0; + + String sequence = seq.toUpperCase(); + + String index = "--------------------------------------&---*---.-----------------@ABCDEFGHIJKLMNOPQRSTUVWXYZ------ABCDEFGHIJKLMNOPQRSTUVWXYZ----@"; + index += "--------------------------------------------------------------------------------------------------------------------------------"; + + for (int i = 0; i < sequence.length(); i++) + { + try + { + if (i < sequence.length()) + { + int pos = index.indexOf(sequence.charAt(i)); + + if (index.charAt(pos)!='_') + { + check += ( ( (i % 57) + 1) * pos); + } + } + } + catch (Exception e) + { + System.err.println("Exception during MSF Checksum calculation"); + e.printStackTrace(); + } } - public static String print(SequenceI[] s) { - StringBuffer out = new StringBuffer("PileUp\n\n"); + return check % 10000; + } - int max = 0; - int maxid = 0; + public static String print(SequenceI[] s) + { + StringBuffer out = new StringBuffer("PileUp\n\n"); - int i = 0; - String big = ""; + int max = 0; + int maxid = 0; - while ((i < s.length) && (s[i] != null)) { - big += s[i].getSequence(); - i++; - } + int i = 0; + String big = ""; - i = 0; + while ( (i < s.length) && (s[i] != null)) + { + big += s[i].getSequence(); + i++; + } - int bigcheck = checkSum(big); + i = 0; - out.append(" MSF: " + s[0].getSequence().length() + - " Type: P Check: " + bigcheck + " ..\n\n\n"); + int bigcheck = checkSum(big); - while ((i < s.length) && (s[i] != null)) { - String seq = s[i].getSequence(); - String name = s[i].getName() + "/" + s[i].getStart() + "-" + - s[i].getEnd(); - int check = checkSum(s[i].getSequence()); - out.append(" Name: " + name + " oo Len: " + - s[i].getSequence().length() + " Check: " + check + - " Weight: 1.00\n"); + out.append(" MSF: " + s[0].getSequence().length() + + " Type: P Check: " + bigcheck + " ..\n\n\n"); - if (seq.length() > max) { - max = seq.length(); - } + while ( (i < s.length) && (s[i] != null)) + { + String seq = s[i].getSequence(); + String name = s[i].getName() + "/" + s[i].getStart() + "-" + + s[i].getEnd(); + int check = checkSum(s[i].getSequence()); + out.append(" Name: " + name + " oo Len: " + + s[i].getSequence().length() + " Check: " + check + + " Weight: 1.00\n"); - if (name.length() > maxid) { - maxid = name.length(); - } + if (seq.length() > max) + { + max = seq.length(); + } - i++; - } + if (name.length() > maxid) + { + maxid = name.length(); + } - if (maxid < 10) { - maxid = 10; - } + i++; + } - maxid++; - out.append("\n\n//\n\n"); + if (maxid < 10) + { + maxid = 10; + } - int len = 50; + maxid++; + out.append("\n\n//\n\n"); - int nochunks = (max / len) + 1; + int len = 50; - if ((max % len) == 0) { - nochunks--; - } + int nochunks = (max / len) + 1; - for (i = 0; i < nochunks; i++) { - int j = 0; - - while ((j < s.length) && (s[j] != null)) { - String name = s[j].getName(); - out.append(new Format("%-" + maxid + "s").form(name + "/" + - s[j].getStart() + "-" + s[j].getEnd()) + " "); - - for (int k = 0; k < 5; k++) { - int start = (i * 50) + (k * 10); - int end = start + 10; - - if ((end < s[j].getSequence().length()) && - (start < s[j].getSequence().length())) { - out.append(s[j].getSequence().substring(start, end)); - - if (k < 4) { - out.append(" "); - } else { - out.append("\n"); - } - } else { - if (start < s[j].getSequence().length()) { - out.append(s[j].getSequence().substring(start)); - out.append("\n"); - } else { - if (k == 0) { - out.append("\n"); - } - } - } - } - - j++; - } + if ( (max % len) == 0) + { + nochunks--; + } - out.append("\n"); + for (i = 0; i < nochunks; i++) + { + int j = 0; + + while ( (j < s.length) && (s[j] != null)) + { + String name = s[j].getName(); + out.append(new Format("%-" + maxid + "s").form(name + "/" + + s[j].getStart() + "-" + s[j].getEnd()) + " "); + + for (int k = 0; k < 5; k++) + { + int start = (i * 50) + (k * 10); + int end = start + 10; + + if ( (end < s[j].getSequence().length()) && + (start < s[j].getSequence().length())) + { + out.append(s[j].getSequence().substring(start, end)); + + if (k < 4) + { + out.append(" "); + } + else + { + out.append("\n"); + } + } + else + { + if (start < s[j].getSequence().length()) + { + out.append(s[j].getSequence().substring(start)); + out.append("\n"); + } + else + { + if (k == 0) + { + out.append("\n"); + } + } + } } - return out.toString(); - } + j++; + } - public String print() { - return print(getSeqsAsArray()); + out.append("\n"); } + + return out.toString(); + } + + public String print() + { + return print(getSeqsAsArray()); + } }