| 1 |
import gnu.getopt.Getopt; |
| 2 |
|
| 3 |
import java.io.BufferedReader; |
| 4 |
import java.io.File; |
| 5 |
import java.io.FileOutputStream; |
| 6 |
import java.io.FileReader; |
| 7 |
import java.io.IOException; |
| 8 |
import java.io.PrintStream; |
| 9 |
import java.sql.SQLException; |
| 10 |
|
| 11 |
import proteinstructure.Pdb; |
| 12 |
import proteinstructure.PdbCodeNotFoundError; |
| 13 |
import proteinstructure.PdbLoadError; |
| 14 |
import proteinstructure.PdbasePdb; |
| 15 |
import tools.MySQLConnection; |
| 16 |
|
| 17 |
|
| 18 |
public class dumpseq { |
| 19 |
/*------------------------------ constants ------------------------------*/ |
| 20 |
|
| 21 |
public static final String PDB_DB = "pdbase"; |
| 22 |
public static final String DB_HOST = "white"; |
| 23 |
public static final String DB_USER = MySQLConnection.getUserName(); |
| 24 |
public static final String DB_PWD = "nieve"; |
| 25 |
|
| 26 |
public static final String GAP_CHARACTER = "-"; |
| 27 |
|
| 28 |
public static void main(String[] args) throws IOException { |
| 29 |
|
| 30 |
|
| 31 |
String help = "Usage, 3 options:\n" + |
| 32 |
"1) genGraph -i <listfile> -o <output_dir> [-D <pdbase_db>] \n" + |
| 33 |
"2) genGraph -p <pdb_code> -c <chain_pdb_code> -o <output_dir> [-D <pdbase_db>] \n" + |
| 34 |
"In case 2) also a list of comma separated pdb codes and chain codes can be specified, e.g. -p 1bxy,1jos -c A,A\n" + |
| 35 |
"If pdbase_db not specified, the default pdbase will be used\n"; |
| 36 |
|
| 37 |
String listfile = ""; |
| 38 |
String[] pdbCodes = null; |
| 39 |
String[] pdbChainCodes = null; |
| 40 |
String pdbaseDb = PDB_DB; |
| 41 |
String outputDir = ""; |
| 42 |
|
| 43 |
Getopt g = new Getopt("genGraph", args, "i:p:c:o:D:h?"); |
| 44 |
int c; |
| 45 |
while ((c = g.getopt()) != -1) { |
| 46 |
switch(c){ |
| 47 |
case 'i': |
| 48 |
listfile = g.getOptarg(); |
| 49 |
break; |
| 50 |
case 'p': |
| 51 |
pdbCodes = g.getOptarg().split(","); |
| 52 |
break; |
| 53 |
case 'c': |
| 54 |
pdbChainCodes = g.getOptarg().split(","); |
| 55 |
break; |
| 56 |
case 'o': |
| 57 |
outputDir = g.getOptarg(); |
| 58 |
break; |
| 59 |
case 'D': |
| 60 |
pdbaseDb = g.getOptarg(); |
| 61 |
break; |
| 62 |
case 'h': |
| 63 |
case '?': |
| 64 |
System.out.println(help); |
| 65 |
System.exit(0); |
| 66 |
break; // getopt() already printed an error |
| 67 |
} |
| 68 |
} |
| 69 |
|
| 70 |
if (listfile.equals("") && pdbCodes==null){ |
| 71 |
System.err.println("Either a listfile or some pdb codes/chain codes must be given"); |
| 72 |
System.err.println(help); |
| 73 |
System.exit(1); |
| 74 |
} |
| 75 |
if (!listfile.equals("") && pdbCodes!=null) { |
| 76 |
System.err.println("Options -p/-c and -i are exclusive. Use only one of them"); |
| 77 |
System.err.println(help); |
| 78 |
System.exit(1); |
| 79 |
} |
| 80 |
|
| 81 |
|
| 82 |
MySQLConnection conn = null; |
| 83 |
|
| 84 |
try{ |
| 85 |
conn = new MySQLConnection(DB_HOST, DB_USER, DB_PWD); |
| 86 |
} catch (Exception e) { |
| 87 |
System.err.println("Error opening database connection. Exiting"); |
| 88 |
System.exit(1); |
| 89 |
} |
| 90 |
|
| 91 |
|
| 92 |
|
| 93 |
|
| 94 |
if (!listfile.equals("")) { |
| 95 |
BufferedReader fpdb = new BufferedReader(new FileReader(listfile)); |
| 96 |
String line = ""; |
| 97 |
int numLines = 0; |
| 98 |
fpdb.mark(100000); |
| 99 |
while ((line = fpdb.readLine() ) != null ) { |
| 100 |
if (line.length()>0) numLines++; |
| 101 |
} |
| 102 |
fpdb.reset(); |
| 103 |
pdbCodes = new String[numLines]; |
| 104 |
pdbChainCodes = new String[numLines]; |
| 105 |
numLines = 0; |
| 106 |
while ((line = fpdb.readLine() ) != null ) { |
| 107 |
pdbCodes[numLines] = line.split("\\s+")[0].toLowerCase(); |
| 108 |
pdbChainCodes[numLines] = line.split("\\s+")[1]; |
| 109 |
numLines++; |
| 110 |
} |
| 111 |
} |
| 112 |
|
| 113 |
int numPdbs = 0; |
| 114 |
|
| 115 |
for (int i=0;i<pdbCodes.length;i++) { |
| 116 |
String pdbCode = pdbCodes[i]; |
| 117 |
String pdbChainCode = pdbChainCodes[i]; |
| 118 |
|
| 119 |
try { |
| 120 |
|
| 121 |
Pdb pdb = new PdbasePdb(pdbCode, pdbaseDb, conn); |
| 122 |
pdb.load(pdbChainCode); |
| 123 |
|
| 124 |
String sequence = pdb.getSequence(); |
| 125 |
|
| 126 |
File outputFile = new File(outputDir,pdbCode+"_"+pdbChainCode+".fasta"); |
| 127 |
|
| 128 |
PrintStream Out = new PrintStream(new FileOutputStream(outputFile.getAbsolutePath())); |
| 129 |
Out.println(">"+pdbCode+"_"+pdbChainCode); |
| 130 |
// for (int pos=1;pos<=sequence.length();pos++) { |
| 131 |
// if (pos%10==0){ |
| 132 |
// Out.printf("%10d",pos/10); |
| 133 |
// } |
| 134 |
// } |
| 135 |
// Out.println(); |
| 136 |
// for (int pos=1;pos<=sequence.length();pos++) { |
| 137 |
// Out.print(pos%10); |
| 138 |
// } |
| 139 |
// Out.println(); |
| 140 |
Out.println(sequence); |
| 141 |
// for (int pos=1;pos<=sequence.length();pos++) { |
| 142 |
// if (pdb.hasCoordinates(pos)) { |
| 143 |
// Out.print(sequence.charAt(pos-1)); |
| 144 |
// } else { |
| 145 |
// Out.print(GAP_CHARACTER); |
| 146 |
// } |
| 147 |
// } |
| 148 |
Out.close(); |
| 149 |
|
| 150 |
System.out.println("Wrote "+outputFile.getAbsolutePath()); |
| 151 |
|
| 152 |
numPdbs++; |
| 153 |
|
| 154 |
} catch (PdbLoadError e) { |
| 155 |
System.err.println("Error loading pdb data for " + pdbCode + pdbChainCode+", specific error: "+e.getMessage()); |
| 156 |
} catch (PdbCodeNotFoundError e) { |
| 157 |
System.err.println("Couldn't find pdb code "+pdbCode); |
| 158 |
} catch (SQLException e) { |
| 159 |
System.err.println("SQL error for structure "+pdbCode+"_"+pdbChainCode+", error: "+e.getMessage()); |
| 160 |
} |
| 161 |
|
| 162 |
} |
| 163 |
|
| 164 |
// output results |
| 165 |
System.out.println("Number of dumped sequences: " + numPdbs); |
| 166 |
|
| 167 |
|
| 168 |
} |
| 169 |
|
| 170 |
|
| 171 |
} |
