/*\r
* Jalview - A Sequence Alignment Editor and Viewer\r
-* Copyright (C) 2005 AM Waterhouse, J Procter, G Barton, M Clamp, S Searle\r
+* Copyright (C) 2006 AM Waterhouse, J Procter, G Barton, M Clamp, S Searle\r
*\r
* This program is free software; you can redistribute it and/or\r
* modify it under the terms of the GNU General Public License\r
* along with this program; if not, write to the Free Software\r
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA\r
*/\r
-\r
package jalview.io;\r
\r
import jalview.datamodel.*;\r
+\r
import jalview.util.*;\r
\r
import java.io.*;\r
+\r
import java.util.*;\r
\r
-public class MSFfile extends AlignFile {\r
\r
- public MSFfile()\r
- {}\r
+/**\r
+ * DOCUMENT ME!\r
+ *\r
+ * @author $author$\r
+ * @version $Revision$\r
+ */\r
+public class MSFfile extends AlignFile\r
+{\r
\r
- public MSFfile(String inStr) {\r
- super(inStr);\r
- }\r
\r
- public MSFfile(String inFile, String type) throws IOException {\r
- super(inFile,type);\r
- }\r
+ /**\r
+ * Creates a new MSFfile object.\r
+ */\r
+ public MSFfile()\r
+ {\r
+ }\r
\r
- private static com.stevesoft.pat.Regex gapre = new com.stevesoft.pat.Regex("\\~","-");\r
- private static com.stevesoft.pat.Regex re2gap = new com.stevesoft.pat.Regex("["+jalview.util.Comparison.GapChars+"]","\\~");\r
\r
- public void parse() {\r
- int i = 0;\r
- boolean seqFlag = false;\r
- String key = new String();\r
- Vector headers = new Vector();\r
- Hashtable seqhash = new Hashtable();\r
- String line;\r
+ /**\r
+ * Creates a new MSFfile object.\r
+ *\r
+ * @param inFile DOCUMENT ME!\r
+ * @param type DOCUMENT ME!\r
+ *\r
+ * @throws IOException DOCUMENT ME!\r
+ */\r
+ public MSFfile(String inFile, String type) throws IOException\r
+ {\r
+ super(inFile, type);\r
+ }\r
\r
- try {\r
- while ((line = nextLine()) != null) {\r
+ /**\r
+ * DOCUMENT ME!\r
+ */\r
+ public void parse() throws IOException\r
+ {\r
+ int i = 0;\r
+ boolean seqFlag = false;\r
+ String key = new String();\r
+ Vector headers = new Vector();\r
+ Hashtable seqhash = new Hashtable();\r
+ String line;\r
+\r
+ try\r
+ {\r
+ while ((line = nextLine()) != null)\r
+ {\r
+ StringTokenizer str = new StringTokenizer(line);\r
+\r
+ while (str.hasMoreTokens())\r
+ {\r
+ String inStr = str.nextToken();\r
+\r
+ //If line has header information add to the headers vector\r
+ if (inStr.indexOf("Name:") != -1)\r
+ {\r
+ key = str.nextToken();\r
+ headers.addElement(key);\r
+ }\r
+\r
+ //if line has // set SeqFlag to 1 so we know sequences are coming\r
+ if (inStr.indexOf("//") != -1)\r
+ {\r
+ seqFlag = true;\r
+ }\r
+\r
+ //Process lines as sequence lines if seqFlag is set\r
+ if ((inStr.indexOf("//") == -1) && (seqFlag == true))\r
+ {\r
+ //seqeunce id is the first field\r
+ key = inStr;\r
+\r
+ StringBuffer tempseq;\r
+\r
+ //Get sequence from hash if it exists\r
+ if (seqhash.containsKey(key))\r
+ {\r
+ tempseq = (StringBuffer) seqhash.get(key);\r
+ }\r
+ else\r
+ {\r
+ tempseq = new StringBuffer();\r
+ seqhash.put(key, tempseq);\r
+ }\r
+\r
+ //loop through the rest of the words\r
+ while (str.hasMoreTokens())\r
+ {\r
+ //append the word to the sequence\r
+ tempseq.append(str.nextToken());\r
+ }\r
+ }\r
+ }\r
+ }\r
+ }\r
+ catch (IOException e)\r
+ {\r
+ System.err.println("Exception parsing MSFFile " + e);\r
+ e.printStackTrace();\r
+ }\r
\r
- StringTokenizer str = new StringTokenizer(line);\r
+ this.noSeqs = headers.size();\r
\r
- while (str.hasMoreTokens()) {\r
+ //Add sequences to the hash\r
+ for (i = 0; i < headers.size(); i++)\r
+ {\r
+ if (seqhash.get(headers.elementAt(i)) != null)\r
+ {\r
+ String head = headers.elementAt(i).toString();\r
+ String seq = seqhash.get(head).toString();\r
\r
- String inStr = str.nextToken();\r
+ if (maxLength < head.length())\r
+ {\r
+ maxLength = head.length();\r
+ }\r
\r
- //If line has header information add to the headers vector\r
- if (inStr.indexOf("Name:") != -1) {\r
- key = str.nextToken();\r
- headers.addElement(key);\r
- }\r
+ // Replace ~ with a sensible gap character\r
+ seq = seq.replace('~', '-');\r
+ if (!isValidProteinSequence(seq.toCharArray()))\r
+ {\r
+ throw new IOException(AppletFormatAdapter.\r
+ INVALID_CHARACTERS\r
+ + " : " + head\r
+ + " : " + invalidCharacter);\r
+ }\r
\r
- //if line has // set SeqFlag to 1 so we know sequences are coming\r
- if (inStr.indexOf("//") != -1) {\r
- seqFlag = true;\r
- }\r
\r
- //Process lines as sequence lines if seqFlag is set\r
- if (( inStr.indexOf("//") == -1) && (seqFlag == true)) {\r
- //seqeunce id is the first field\r
- key = inStr;\r
- StringBuffer tempseq;\r
-\r
- //Get sequence from hash if it exists\r
- if (seqhash.containsKey(key)) {\r
- tempseq = (StringBuffer)seqhash.get(key);\r
- } else {\r
- tempseq = new StringBuffer();\r
- seqhash.put(key,tempseq);\r
- }\r
-\r
- //loop through the rest of the words\r
- while (str.hasMoreTokens()) {\r
- //append the word to the sequence\r
- tempseq.append(str.nextToken());\r
- }\r
+ Sequence newSeq = parseId(head);\r
+\r
+ newSeq.setSequence(seq);\r
+\r
+ seqs.addElement(newSeq);\r
+ }\r
+ else\r
+ {\r
+ System.err.println("MSFFile Parser: Can't find sequence for " +\r
+ headers.elementAt(i));\r
+ }\r
}\r
- }\r
}\r
- } catch (IOException e) {\r
- System.err.println("Exception parsing MSFFile " + e);\r
- e.printStackTrace();\r
+\r
+ /**\r
+ * DOCUMENT ME!\r
+ *\r
+ * @param seq DOCUMENT ME!\r
+ *\r
+ * @return DOCUMENT ME!\r
+ */\r
+ public int checkSum(String seq)\r
+ {\r
+ int check = 0;\r
+ String sequence = seq.toUpperCase();\r
+\r
+ for (int i = 0; i < sequence.length(); i++)\r
+ {\r
+ try\r
+ {\r
+\r
+ int value = sequence.charAt(i);\r
+ if (value!=-1)\r
+ {\r
+ check += (i % 57 +1) * value;\r
+ }\r
+ }\r
+ catch (Exception e)\r
+ {\r
+ System.err.println("Exception during MSF Checksum calculation");\r
+ e.printStackTrace();\r
+ }\r
+ }\r
+\r
+ return check % 10000;\r
}\r
\r
- this.noSeqs = headers.size();\r
\r
- //Add sequences to the hash\r
- for (i = 0; i < headers.size(); i++ ) {\r
+ /**\r
+ * DOCUMENT ME!\r
+ *\r
+ * @param s DOCUMENT ME!\r
+ * @param is_NA DOCUMENT ME!\r
+ *\r
+ * @return DOCUMENT ME!\r
+ */\r
+ public String print(SequenceI[] seqs)\r
+ {\r
\r
- if ( seqhash.get(headers.elementAt(i)) != null) {\r
- String head = headers.elementAt(i).toString();\r
- String seq = seqhash.get(head).toString();\r
+ boolean is_NA = jalview.util.Comparison.isNucleotide(seqs);\r
\r
- int start = 1;\r
- int end = seq.length();\r
+ SequenceI [] s = new SequenceI[seqs.length];\r
\r
- if (maxLength < head.length() ) {\r
- maxLength = head.length();\r
- }\r
+ StringBuffer out = new StringBuffer("!!" + (is_NA ? "NA" : "AA") +\r
+ "_MULTIPLE_ALIGNMENT 1.0\n\n"); // TODO: JBPNote : Jalview doesn't remember NA or AA yet.\r
\r
- if (head.indexOf("/") > 0 ) {\r
+ int max = 0;\r
+ int maxid = 0;\r
+ int i = 0;\r
\r
- StringTokenizer st = new StringTokenizer(head,"/");\r
+ while ((i < seqs.length) && (seqs[i] != null))\r
+ {\r
+ // Replace all internal gaps with . and external spaces with ~\r
+ s[i] =new Sequence(seqs[i].getName(),seqs[i].getSequenceAsString().replace('-', '.'));\r
\r
- if (st.countTokens() == 2) {\r
+ StringBuffer sb = new StringBuffer();\r
+ sb.append(s[i].getSequence());\r
\r
- head = st.nextToken();\r
- String tmp = st.nextToken();\r
- st = new StringTokenizer(tmp,"-");\r
- if (st.countTokens() == 2) {\r
- start = Integer.valueOf(st.nextToken()).intValue();\r
- end = Integer.valueOf(st.nextToken()).intValue();\r
+ for (int ii = 0; ii < sb.length(); ii++)\r
+ {\r
+ if (sb.charAt(ii) == '.')\r
+ {\r
+ sb.setCharAt(ii, '~');\r
}\r
+ else\r
+ break;\r
}\r
- }\r
- // Replace ~ with a sensible gap character\r
- seq = gapre.replaceAll(seq);\r
- Sequence newSeq = new Sequence(head,seq,start,end);\r
\r
- seqs.addElement(newSeq);\r
-\r
- } else {\r
- System.err.println("MSFFile Parser: Can't find sequence for " + headers.elementAt(i));\r
- }\r
- }\r
+ for (int ii = sb.length() - 1; ii > 0; ii--)\r
+ {\r
+ if (sb.charAt(ii) == '.')\r
+ {\r
+ sb.setCharAt(ii,'~');\r
+ }\r
+ else\r
+ break;\r
+ }\r
\r
- }\r
+ s[i].setSequence(sb.toString());\r
\r
- public static int checkSum(String seq) {\r
- //String chars = "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz.*~&@";\r
- int check = 0;\r
+ if (s[i].getSequence().length > max)\r
+ {\r
+ max = s[i].getSequence().length;\r
+ }\r
\r
- String index = "--------------------------------------&---*---.-----------------@ABCDEFGHIJKLMNOPQRSTUVWXYZ------ABCDEFGHIJKLMNOPQRSTUVWXYZ----@";\r
- index += "--------------------------------------------------------------------------------------------------------------------------------";\r
+ i++;\r
+ }\r
\r
- for(int i = 0; i < seq.length(); i++) {\r
- try {\r
- if (i <seq.length()) {\r
- int pos = index.indexOf(seq.substring(i,i+1));\r
- if (!index.substring(pos,pos+1).equals("_")) {\r
- check += ((i % 57) + 1) * pos;\r
- }\r
+ Format maxLenpad = new Format("%" + (new String("" + max)).length() +\r
+ "d");\r
+ Format maxChkpad = new Format("%" + (new String("1" + max)).length() +\r
+ "d");\r
+ i = 0;\r
+\r
+ int bigChecksum = 0;\r
+ int [] checksums = new int[s.length];\r
+ while ( i < s.length )\r
+ {\r
+ checksums[i] = checkSum(s[i].getSequenceAsString());\r
+ bigChecksum += checksums[i];\r
+ i++;\r
}\r
- } catch (Exception e) {\r
- System.err.println("Exception during MSF Checksum calculation");\r
- e.printStackTrace();\r
- }\r
- }\r
- return check % 10000;\r
- }\r
-\r
- public static String print(SequenceI[] s) {\r
- return print(s, false);\r
- }\r
- public static String print(SequenceI[] s, boolean is_NA) {\r
- 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
-\r
- int max = 0;\r
- int maxid = 0;\r
- int i = 0;\r
- String big = "";\r
-\r
- while (i < s.length && s[i] != null) {\r
- String sq;\r
- big += (sq=s[i].getSequence());\r
- if (sq.length() > max) {\r
- max = sq.length();\r
- }\r
- i++;\r
- }\r
- Format maxLenpad = new Format("%"+(new String(""+max)).length()+"d");\r
- Format maxChkpad = new Format("%"+(new String("1"+max)).length()+"d");\r
- i = 0;\r
- long bigcheck = checkSum(big);\r
- long maxNB=0;\r
- out.append(" MSF: " + s[0].getSequence().length() + " Type: "+(is_NA?"N":"P")+" Check: " + bigcheck + " ..\n\n\n");\r
- String nameBlock[] = new String[s.length];\r
- String idBlock[] = new String[s.length];\r
- while (i < s.length && s[i] != null) {\r
- String seq = s[i].getSequence();\r
- String name = s[i].getName()+ "/" + s[i].getStart() + "-" + s[i].getEnd();\r
- int check = checkSum(s[i].getSequence());\r
- nameBlock[i]=new String(" Name: "+name+" ");\r
- idBlock[i] = new String("Len: " + maxLenpad.form(s[i].getSequence().length()) + " Check:" + maxChkpad.form(check) + " Weight: 1.00\n");\r
-\r
- if (name.length() > maxid) {\r
- maxid = name.length();\r
- }\r
- if (nameBlock[i].length()>maxNB) {\r
- maxNB=nameBlock[i].length();\r
- }\r
-\r
- i++;\r
- }\r
- if (maxid < 10) {\r
- maxid = 10;\r
- }\r
- if (maxNB<15) {\r
- maxNB=15;\r
- }\r
- Format nbFormat = new Format("%-"+maxNB+"s");\r
- for (i=0;i<s.length && s[i]!=null;i++) {\r
- out.append(nbFormat.form(nameBlock[i])+idBlock[i]);\r
- }\r
- maxid++;\r
- out.append( "\n\n//\n\n");\r
\r
- int len = 50;\r
+ long maxNB = 0;\r
+ out.append(" MSF: " + s[0].getSequence().length + " Type: " +\r
+ (is_NA ? "N" : "P") + " Check: " + (bigChecksum%10000) + " ..\n\n\n");\r
\r
- int nochunks = max / len + 1;\r
- if (max%len == 0) {\r
- nochunks--;\r
- }\r
- for (i = 0; i < nochunks; i++) {\r
- int j = 0;\r
- while (j < s.length && s[j] != null) {\r
- String name = s[j].getName();\r
- out.append( new Format("%-" + maxid + "s").form(name + "/" + s[j].getStart() + "-" + s[j].getEnd()) + " ");\r
- for (int k = 0; k < 5; k++) {\r
-\r
- int start = i*50 + k*10;\r
- int end = start + 10;\r
-\r
- if (end < s[j].getSequence().length() && start < s[j].getSequence().length() ) {\r
- out.append(re2gap.replaceAll(s[j].getSequence().substring(start,end)));\r
- if (k < 4) {\r
- // out.append(" ");\r
- } else {\r
- out.append("\n");\r
+ String[] nameBlock = new String[s.length];\r
+ String[] idBlock = new String[s.length];\r
+\r
+ i=0;\r
+ while ((i < s.length) && (s[i] != null))\r
+ {\r
+\r
+ nameBlock[i] = new String(" Name: " + printId(s[i])+" ");\r
+\r
+ idBlock[i] = new String("Len: " +\r
+ maxLenpad.form(s[i].getSequence().length) + " Check: " +\r
+ maxChkpad.form(checksums[i]) + " Weight: 1.00\n");\r
+\r
+ if (s[i].getName().length() > maxid)\r
+ {\r
+ maxid = s[i].getName().length();\r
}\r
- } else {\r
- if (start < s[j].getSequence().length()) {\r
- out.append(re2gap.replaceAll(s[j].getSequence().substring(start)));\r
- out.append("\n");\r
- } else {\r
- if (k == 0) {\r
- out.append("\n");\r
- }\r
+\r
+ if (nameBlock[i].length() > maxNB)\r
+ {\r
+ maxNB = nameBlock[i].length();\r
}\r
- }\r
+\r
+ i++;\r
}\r
- j++;\r
- }\r
- out.append("\n");\r
\r
- }\r
- return out.toString();\r
- }\r
- public String print() {\r
- return print(getSeqsAsArray());\r
- }\r
-}\r
+ if (maxid < 10)\r
+ {\r
+ maxid = 10;\r
+ }\r
+\r
+ if (maxNB < 15)\r
+ {\r
+ maxNB = 15;\r
+ }\r
+\r
+ Format nbFormat = new Format("%-" + maxNB + "s");\r
+\r
+ for (i = 0; (i < s.length) && (s[i] != null); i++)\r
+ {\r
+ out.append(nbFormat.form(nameBlock[i]) + idBlock[i]);\r
+ }\r
\r
+ maxid++;\r
+ out.append("\n\n//\n\n");\r
\r
+ int len = 50;\r
\r
+ int nochunks = (max / len) + 1;\r
\r
+ if ((max % len) == 0)\r
+ {\r
+ nochunks--;\r
+ }\r
\r
+ for (i = 0; i < nochunks; i++)\r
+ {\r
+ int j = 0;\r
+\r
+ while ((j < s.length) && (s[j] != null))\r
+ {\r
+ String name = printId( s[j] );\r
+\r
+ out.append(new Format("%-" + maxid + "s").form(name+" "));\r
+\r
+\r
+ for (int k = 0; k < 5; k++)\r
+ {\r
+ int start = (i * 50) + (k * 10);\r
+ int end = start + 10;\r
+\r
+ if ((end < s[j].getSequence().length) &&\r
+ (start < s[j].getSequence().length))\r
+ {\r
+ out.append(s[j].getSequence(start, end));\r
+\r
+ if (k < 4)\r
+ {\r
+ out.append(" ");\r
+ }\r
+ else\r
+ {\r
+ out.append("\n");\r
+ }\r
+ }\r
+ else\r
+ {\r
+ if (start < s[j].getSequence().length)\r
+ {\r
+ out.append(s[j].getSequenceAsString().substring(start));\r
+ out.append("\n");\r
+ }\r
+ else\r
+ {\r
+ if (k == 0)\r
+ {\r
+ out.append("\n");\r
+ }\r
+ }\r
+ }\r
+ }\r
+\r
+ j++;\r
+ }\r
\r
+ out.append("\n");\r
+ }\r
\r
+ return out.toString();\r
+ }\r
+\r
+ /**\r
+ * DOCUMENT ME!\r
+ *\r
+ * @return DOCUMENT ME!\r
+ */\r
+ public String print()\r
+ {\r
+ return print(getSeqsAsArray());\r
+ }\r
+}\r
git://source.jalview.org / jalview.git / blobdiff