* 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
+\r
public class MSFfile extends AlignFile {\r
+ private static com.stevesoft.pat.Regex gapre = new com.stevesoft.pat.Regex("\\~",\r
+ "-");\r
+ private static com.stevesoft.pat.Regex re2gap = new com.stevesoft.pat.Regex(\r
+ "[" + jalview.util.Comparison.GapChars + "]", "\\~");\r
\r
- public MSFfile()\r
- {}\r
+ public MSFfile() {\r
+ }\r
\r
- public MSFfile(String inStr) {\r
- super(inStr);\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
+ public MSFfile(String inFile, String type) throws IOException {\r
+ super(inFile, type);\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
+ 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
+ try {\r
+ while ((line = nextLine()) != null) {\r
+ StringTokenizer str = new StringTokenizer(line);\r
+\r
+ while (str.hasMoreTokens()) {\r
+ String inStr = str.nextToken();\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
+\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
+\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
+ }\r
+ }\r
+ }\r
+ } catch (IOException e) {\r
+ System.err.println("Exception parsing MSFFile " + e);\r
+ e.printStackTrace();\r
+ }\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
+ this.noSeqs = headers.size();\r
\r
- try {\r
- while ((line = nextLine()) != null) {\r
+ //Add sequences to the hash\r
+ for (i = 0; i < headers.size(); i++) {\r
+ if (seqhash.get(headers.elementAt(i)) != null) {\r
+ String head = headers.elementAt(i).toString();\r
+ String seq = seqhash.get(head).toString();\r
\r
- StringTokenizer str = new StringTokenizer(line);\r
+ int start = 1;\r
+ int end = seq.length();\r
\r
- while (str.hasMoreTokens()) {\r
+ if (maxLength < head.length()) {\r
+ maxLength = head.length();\r
+ }\r
\r
- String inStr = str.nextToken();\r
+ if (head.indexOf("/") > 0) {\r
+ StringTokenizer st = new StringTokenizer(head, "/");\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
+ if (st.countTokens() == 2) {\r
+ head = st.nextToken();\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
+ String tmp = st.nextToken();\r
+ st = new StringTokenizer(tmp, "-");\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
- }\r
- }\r
- }\r
- } catch (IOException e) {\r
- System.err.println("Exception parsing MSFFile " + e);\r
- e.printStackTrace();\r
- }\r
-\r
- this.noSeqs = headers.size();\r
+ if (st.countTokens() == 2) {\r
+ start = Integer.valueOf(st.nextToken()).intValue();\r
+ end = Integer.valueOf(st.nextToken()).intValue();\r
+ }\r
+ }\r
+ }\r
\r
- //Add sequences to the hash\r
- for (i = 0; i < headers.size(); i++ ) {\r
+ // Replace ~ with a sensible gap character\r
+ seq = gapre.replaceAll(seq);\r
\r
- if ( seqhash.get(headers.elementAt(i)) != null) {\r
- String head = headers.elementAt(i).toString();\r
- String seq = seqhash.get(head).toString();\r
+ Sequence newSeq = new Sequence(head, seq, start, end);\r
\r
- int start = 1;\r
- int end = seq.length();\r
+ seqs.addElement(newSeq);\r
+ } else {\r
+ System.err.println("MSFFile Parser: Can't find sequence for " +\r
+ headers.elementAt(i));\r
+ }\r
+ }\r
+ }\r
\r
- if (maxLength < head.length() ) {\r
- maxLength = head.length();\r
+ public static int checkSum(String seq) {\r
+ //String chars = "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz.*~&@";\r
+ int check = 0;\r
+\r
+ String index = "--------------------------------------&---*---.-----------------@ABCDEFGHIJKLMNOPQRSTUVWXYZ------ABCDEFGHIJKLMNOPQRSTUVWXYZ----@";\r
+ index += "--------------------------------------------------------------------------------------------------------------------------------";\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
+\r
+ if (!index.substring(pos, pos + 1).equals("_")) {\r
+ check += (((i % 57) + 1) * pos);\r
+ }\r
+ }\r
+ } catch (Exception e) {\r
+ System.err.println("Exception during MSF Checksum calculation");\r
+ e.printStackTrace();\r
+ }\r
}\r
\r
- if (head.indexOf("/") > 0 ) {\r
+ return check % 10000;\r
+ }\r
+\r
+ public static String print(SequenceI[] s) {\r
+ return print(s, false);\r
+ }\r
+\r
+ public static String print(SequenceI[] s, boolean is_NA) {\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
- StringTokenizer st = new StringTokenizer(head,"/");\r
+ int max = 0;\r
+ int maxid = 0;\r
+ int i = 0;\r
+ String big = "";\r
\r
- if (st.countTokens() == 2) {\r
+ while ((i < s.length) && (s[i] != null)) {\r
+ String sq;\r
+ big += (sq = 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
+ if (sq.length() > max) {\r
+ max = sq.length();\r
}\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
+ i++;\r
+ }\r
\r
- } else {\r
- System.err.println("MSFFile Parser: Can't find sequence for " + headers.elementAt(i));\r
- }\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
+ long bigcheck = checkSum(big);\r
+ long maxNB = 0;\r
+ out.append(" MSF: " + s[0].getSequence().length() + " Type: " +\r
+ (is_NA ? "N" : "P") + " Check: " + bigcheck + " ..\n\n\n");\r
+\r
+ String[] nameBlock = new String[s.length];\r
+ String[] idBlock = new String[s.length];\r
+\r
+ while ((i < s.length) && (s[i] != null)) {\r
+ String seq = s[i].getSequence();\r
+ String name = s[i].getName() + "/" + s[i].getStart() + "-" +\r
+ s[i].getEnd();\r
+ int check = checkSum(s[i].getSequence());\r
+ nameBlock[i] = new String(" Name: " + name + " ");\r
+ idBlock[i] = new String("Len: " +\r
+ maxLenpad.form(s[i].getSequence().length()) + " Check:" +\r
+ maxChkpad.form(check) + " Weight: 1.00\n");\r
+\r
+ if (name.length() > maxid) {\r
+ maxid = name.length();\r
+ }\r
\r
- }\r
+ if (nameBlock[i].length() > maxNB) {\r
+ maxNB = nameBlock[i].length();\r
+ }\r
\r
- public static int checkSum(String seq) {\r
- //String chars = "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz.*~&@";\r
- int check = 0;\r
+ i++;\r
+ }\r
\r
- String index = "--------------------------------------&---*---.-----------------@ABCDEFGHIJKLMNOPQRSTUVWXYZ------ABCDEFGHIJKLMNOPQRSTUVWXYZ----@";\r
- index += "--------------------------------------------------------------------------------------------------------------------------------";\r
+ if (maxid < 10) {\r
+ maxid = 10;\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
+ if (maxNB < 15) {\r
+ maxNB = 15;\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
+ Format nbFormat = new Format("%-" + maxNB + "s");\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
- }\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
- }\r
+ for (i = 0; (i < s.length) && (s[i] != null); i++) {\r
+ out.append(nbFormat.form(nameBlock[i]) + idBlock[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
+ 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
+ nochunks--;\r
+ }\r
\r
+ for (i = 0; i < nochunks; i++) {\r
+ int j = 0;\r
+\r
+ while ((j < s.length) && (s[j] != null)) {\r
+ String name = s[j].getName();\r
+ out.append(new Format("%-" + maxid + "s").form(name + "/" +\r
+ s[j].getStart() + "-" + s[j].getEnd()) + " ");\r
+\r
+ for (int k = 0; k < 5; k++) {\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
+ out.append(re2gap.replaceAll(s[j].getSequence()\r
+ .substring(start, end)));\r
+\r
+ if (k < 4) {\r
+ // out.append(" ");\r
+ } else {\r
+ out.append("\n");\r
+ }\r
+ } else {\r
+ if (start < s[j].getSequence().length()) {\r
+ out.append(re2gap.replaceAll(\r
+ s[j].getSequence().substring(start)));\r
+ out.append("\n");\r
+ } else {\r
+ if (k == 0) {\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
+ public String print() {\r
+ return print(getSeqsAsArray());\r
+ }\r
+}\r
git://source.jalview.org / jalview.git / blobdiff