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.*; - - -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); - } - - //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(); - } - - 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(); - - if (maxLength < head.length()) { - maxLength = head.length(); - } - - if (head.indexOf("/") > 0) { - StringTokenizer st = new StringTokenizer(head, "/"); - - if (st.countTokens() == 2) { - head = st.nextToken(); - - String tmp = st.nextToken(); - st = new StringTokenizer(tmp, "-"); - - if (st.countTokens() == 2) { - start = Integer.valueOf(st.nextToken()).intValue(); - end = Integer.valueOf(st.nextToken()).intValue(); - } - } - } - - Sequence newSeq = new Sequence(head, seq, start, end); - - 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 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(); - } - } - - return check % 10000; - } - - public static String print(SequenceI[] s) { - StringBuffer out = new StringBuffer("PileUp\n\n"); - - int max = 0; - int maxid = 0; - - int i = 0; - String big = ""; - - while ((i < s.length) && (s[i] != null)) { - big += s[i].getSequence(); - i++; - } - - i = 0; - - int bigcheck = checkSum(big); - - out.append(" MSF: " + s[0].getSequence().length() + - " Type: P Check: " + bigcheck + " ..\n\n\n"); - - 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 (seq.length() > max) { - max = seq.length(); - } - - if (name.length() > maxid) { - maxid = name.length(); - } - - i++; - } - - if (maxid < 10) { - maxid = 10; - } - - maxid++; - out.append("\n\n//\n\n"); - - int len = 50; - - int nochunks = (max / len) + 1; - - if ((max % len) == 0) { - nochunks--; - } - - 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++; - } - - out.append("\n"); - } - - return out.toString(); - } - - public String print() { - return print(getSeqsAsArray()); - } -} + *+ * 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.SequenceI; +import jalview.util.Format; + +import java.io.IOException; + +public class PileUpfile extends MSFfile +{ + + /** + * Creates a new MSFfile object. + */ + public PileUpfile() + { + } + + /** + * Creates a new MSFfile object. + * + * @param inFile + * DOCUMENT ME! + * @param type + * DOCUMENT ME! + * + * @throws IOException + * DOCUMENT ME! + */ + public PileUpfile(String inFile, String type) throws IOException + { + super(inFile, type); + } + + public PileUpfile(FileParse source) throws IOException + { + super(source); + } + + /** + * DOCUMENT ME! + * + * @return DOCUMENT ME! + */ + public String print() + { + return print(getSeqsAsArray()); + } + + public String print(SequenceI[] s) + { + StringBuffer out = new StringBuffer("PileUp"); + out.append(newline); + out.append(newline); + + int max = 0; + int maxid = 0; + + int i = 0; + int bigChecksum = 0; + int[] checksums = new int[s.length]; + while (i < s.length) + { + checksums[i] = checkSum(s[i].getSequenceAsString()); + bigChecksum += checksums[i]; + i++; + } + + out.append(" MSF: " + s[0].getSequence().length + + " Type: P Check: " + bigChecksum % 10000 + " .."); + out.append(newline); + out.append(newline); + out.append(newline); + + i = 0; + while ((i < s.length) && (s[i] != null)) + { + String seq = s[i].getSequenceAsString(); + out.append(" Name: " + printId(s[i]) + " oo Len: " + seq.length() + + " Check: " + checksums[i] + " Weight: 1.00"); + out.append(newline); + + if (seq.length() > max) + { + max = seq.length(); + } + + if (s[i].getName().length() > maxid) + { + maxid = s[i].getName().length(); + } + + i++; + } + + if (maxid < 10) + { + maxid = 10; + } + + maxid++; + out.append(newline); + out.append(newline); + out.append("//"); + out.append(newline); + out.append(newline); + + int len = 50; + + int nochunks = (max / len) + 1; + + if ((max % len) == 0) + { + nochunks--; + } + + for (i = 0; i < nochunks; i++) + { + int j = 0; + + while ((j < s.length) && (s[j] != null)) + { + String name = printId(s[j]); + + out.append(new Format("%-" + maxid + "s").form(name + " ")); + + 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(start, end)); + + if (k < 4) + { + out.append(" "); + } + else + { + out.append(newline); + } + } + else + { + if (start < s[j].getSequence().length) + { + out.append(s[j].getSequenceAsString().substring(start)); + out.append(newline); + } + else + { + if (k == 0) + { + out.append(newline); + } + } + } + } + + j++; + } + + out.append(newline); + } + + return out.toString(); + } +}