1 |
duarte |
277 |
import gnu.getopt.Getopt; |
2 |
|
|
|
3 |
|
|
import java.io.BufferedReader; |
4 |
|
|
import java.io.File; |
5 |
|
|
import java.io.FileReader; |
6 |
|
|
import java.io.IOException; |
7 |
|
|
import java.sql.SQLException; |
8 |
|
|
|
9 |
|
|
|
10 |
|
|
import proteinstructure.Graph; |
11 |
|
|
import proteinstructure.Pdb; |
12 |
|
|
import proteinstructure.PdbChainCodeNotFoundError; |
13 |
|
|
import proteinstructure.PdbCodeNotFoundError; |
14 |
|
|
import proteinstructure.PdbaseInconsistencyError; |
15 |
|
|
import proteinstructure.PdbasePdb; |
16 |
|
|
import proteinstructure.PdbfileFormatError; |
17 |
|
|
import proteinstructure.PdbfilePdb; |
18 |
|
|
import tools.MySQLConnection; |
19 |
|
|
|
20 |
|
|
|
21 |
|
|
public class genGraph { |
22 |
|
|
/*------------------------------ constants ------------------------------*/ |
23 |
|
|
|
24 |
|
|
public static final String PDB_DB = "pdbase"; |
25 |
|
|
public static final String DB_HOST = "white"; |
26 |
|
|
public static final String DB_USER = getUserName(); |
27 |
|
|
public static final String DB_PWD = "nieve"; |
28 |
|
|
public static final String DSSP_EXE = "/project/StruPPi/bin/dssp"; |
29 |
|
|
public static final String DSSP_PARAMS = "--"; |
30 |
|
|
|
31 |
|
|
//public static double cutoff = 4.2; |
32 |
|
|
//public static String edgeType = "ALL"; |
33 |
|
|
|
34 |
|
|
/*---------------------------- private methods --------------------------*/ |
35 |
|
|
/** |
36 |
|
|
* Get user name from operating system (for use as database username). |
37 |
|
|
* */ |
38 |
|
|
private static String getUserName() { |
39 |
|
|
String user = null; |
40 |
|
|
user = System.getProperty("user.name"); |
41 |
|
|
if(user == null) { |
42 |
|
|
System.err.println("Could not get user name from operating system."); |
43 |
|
|
} |
44 |
|
|
return user; |
45 |
|
|
} |
46 |
|
|
|
47 |
|
|
public static void main(String[] args) throws IOException { |
48 |
|
|
|
49 |
|
|
|
50 |
|
|
String help = "Usage, 3 options:\n" + |
51 |
|
|
"1) genGraph -i <listfile> -d <distance_cutoff> -t <contact_type> -o <output_dir> [-D <pdbase_db>] \n" + |
52 |
|
|
"2) genGraph -p <pdb_code> -c <chain_pdb_code> -d <distance_cutoff> -t <contact_type> -o <output_dir> [-D <pdbase_db>] \n" + |
53 |
|
|
"3) genGraph -f <pdbfile> -c <chain_pdb_code> -d <distance_cutoff> -t <contact_type> -o <output_dir> \n" + |
54 |
|
|
"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" + |
55 |
|
|
"If pdbase_db not specified, the default pdbase will be used\n"; |
56 |
|
|
|
57 |
|
|
String listfile = ""; |
58 |
|
|
String[] pdbCodes = null; |
59 |
|
|
String[] pdbChainCodes = null; |
60 |
|
|
String pdbfile = ""; |
61 |
|
|
String pdbaseDb = PDB_DB; |
62 |
|
|
String edgeType = ""; |
63 |
|
|
double cutoff = 0.0; |
64 |
|
|
String outputDir = ""; |
65 |
|
|
|
66 |
|
|
Getopt g = new Getopt("genGraph", args, "i:p:c:f:d:t:o:D:h?"); |
67 |
|
|
int c; |
68 |
|
|
while ((c = g.getopt()) != -1) { |
69 |
|
|
switch(c){ |
70 |
|
|
case 'i': |
71 |
|
|
listfile = g.getOptarg(); |
72 |
|
|
break; |
73 |
|
|
case 'p': |
74 |
|
|
pdbCodes = g.getOptarg().split(","); |
75 |
|
|
break; |
76 |
|
|
case 'c': |
77 |
|
|
pdbChainCodes = g.getOptarg().split(","); |
78 |
|
|
break; |
79 |
|
|
case 'f': |
80 |
|
|
pdbfile = g.getOptarg(); |
81 |
|
|
break; |
82 |
|
|
case 'd': |
83 |
|
|
cutoff = Double.valueOf(g.getOptarg()); |
84 |
|
|
break; |
85 |
|
|
case 't': |
86 |
|
|
edgeType = g.getOptarg(); |
87 |
|
|
break; |
88 |
|
|
case 'o': |
89 |
|
|
outputDir = g.getOptarg(); |
90 |
|
|
break; |
91 |
|
|
case 'D': |
92 |
|
|
pdbaseDb = g.getOptarg(); |
93 |
|
|
break; |
94 |
|
|
case 'h': |
95 |
|
|
case '?': |
96 |
|
|
System.out.println(help); |
97 |
|
|
System.exit(0); |
98 |
|
|
break; // getopt() already printed an error |
99 |
|
|
} |
100 |
|
|
} |
101 |
|
|
|
102 |
|
|
if (outputDir.equals("") || edgeType.equals("") || cutoff==0.0) { |
103 |
|
|
System.err.println("Some missing option"); |
104 |
|
|
System.err.println(help); |
105 |
|
|
System.exit(1); |
106 |
|
|
} |
107 |
|
|
if (listfile.equals("") && pdbCodes==null && pdbfile.equals("")){ |
108 |
|
|
System.err.println("Either a listfile, some pdb codes/chain codes or a pdbfile must be given"); |
109 |
|
|
System.err.println(help); |
110 |
|
|
System.exit(1); |
111 |
|
|
} |
112 |
|
|
if ((!listfile.equals("") && pdbCodes!=null) || (!listfile.equals("") && !pdbfile.equals("")) || (pdbCodes!=null && !pdbfile.equals(""))) { |
113 |
|
|
System.err.println("Options -p/-c, -i and -f/-c are exclusive. Use only one of them"); |
114 |
|
|
System.err.println(help); |
115 |
|
|
System.exit(1); |
116 |
|
|
} |
117 |
|
|
|
118 |
|
|
|
119 |
|
|
MySQLConnection conn = null; |
120 |
|
|
|
121 |
|
|
try{ |
122 |
|
|
conn = new MySQLConnection(DB_HOST, DB_USER, DB_PWD); |
123 |
|
|
} catch (Exception e) { |
124 |
|
|
System.err.println("Error opening database connection. Exiting"); |
125 |
|
|
System.exit(1); |
126 |
|
|
} |
127 |
|
|
|
128 |
|
|
|
129 |
|
|
if (pdbfile.equals("")){ |
130 |
|
|
|
131 |
|
|
if (!listfile.equals("")) { |
132 |
|
|
BufferedReader fpdb = new BufferedReader(new FileReader(listfile)); |
133 |
|
|
String line = ""; |
134 |
|
|
int numLines = 0; |
135 |
|
|
fpdb.mark(100000); |
136 |
|
|
while ((line = fpdb.readLine() ) != null ) { |
137 |
duarte |
285 |
if (line.length()>0) numLines++; |
138 |
duarte |
277 |
} |
139 |
|
|
fpdb.reset(); |
140 |
|
|
pdbCodes = new String[numLines]; |
141 |
|
|
pdbChainCodes = new String[numLines]; |
142 |
|
|
numLines = 0; |
143 |
|
|
while ((line = fpdb.readLine() ) != null ) { |
144 |
duarte |
285 |
pdbCodes[numLines] = line.split("\\s+")[0].toLowerCase(); |
145 |
|
|
pdbChainCodes[numLines] = line.split("\\s+")[1]; |
146 |
duarte |
277 |
numLines++; |
147 |
|
|
} |
148 |
|
|
} |
149 |
|
|
|
150 |
|
|
int numPdbs = 0; |
151 |
|
|
|
152 |
|
|
for (int i=0;i<pdbCodes.length;i++) { |
153 |
|
|
String pdbCode = pdbCodes[i]; |
154 |
|
|
String pdbChainCode = pdbChainCodes[i]; |
155 |
|
|
|
156 |
|
|
try { |
157 |
duarte |
285 |
|
158 |
|
|
long start = System.currentTimeMillis(); |
159 |
|
|
|
160 |
duarte |
277 |
Pdb pdb = new PdbasePdb(pdbCode, pdbChainCode, pdbaseDb, conn); |
161 |
|
|
|
162 |
|
|
// get graph |
163 |
|
|
Graph graph = pdb.get_graph(edgeType, cutoff); |
164 |
|
|
|
165 |
|
|
File outputFile = new File(outputDir,pdbCode+"_"+pdbChainCode+"_"+edgeType+"_"+cutoff+".graph"); |
166 |
|
|
graph.write_graph_to_file(outputFile.getAbsolutePath()); |
167 |
|
|
|
168 |
duarte |
285 |
long end = System.currentTimeMillis(); |
169 |
|
|
double time = (double) (end -start)/1000; |
170 |
|
|
|
171 |
duarte |
277 |
System.out.println("Wrote "+outputFile.getAbsolutePath()); |
172 |
duarte |
285 |
System.out.printf("%5.3f s\n",time); |
173 |
|
|
|
174 |
duarte |
277 |
numPdbs++; |
175 |
|
|
|
176 |
|
|
} catch (PdbaseInconsistencyError e) { |
177 |
|
|
System.err.println("Inconsistency in " + pdbCode + pdbChainCode); |
178 |
|
|
} catch (PdbCodeNotFoundError e) { |
179 |
|
|
System.err.println("Couldn't find pdb code "+pdbCode); |
180 |
|
|
} catch (SQLException e) { |
181 |
duarte |
285 |
System.err.println("SQL error for structure "+pdbCode+"_"+pdbChainCode+", error: "+e.getMessage()); |
182 |
duarte |
277 |
} catch (PdbChainCodeNotFoundError e) { |
183 |
|
|
System.err.println("Couldn't find pdb chain code "+pdbChainCode+" for pdb code "+pdbCode); |
184 |
|
|
} |
185 |
|
|
|
186 |
|
|
} |
187 |
|
|
|
188 |
|
|
// output results |
189 |
|
|
System.out.println("Number of structures done successfully: " + numPdbs); |
190 |
|
|
|
191 |
|
|
|
192 |
|
|
} else { |
193 |
|
|
String pdbChainCode = pdbChainCodes[0]; |
194 |
|
|
try { |
195 |
|
|
Pdb pdb = new PdbfilePdb(pdbfile,pdbChainCode); |
196 |
|
|
if (!pdb.hasSecondaryStructure()) { |
197 |
|
|
pdb.runDssp(DSSP_EXE, DSSP_PARAMS); |
198 |
|
|
} |
199 |
|
|
Graph graph = pdb.get_graph(edgeType, cutoff); |
200 |
|
|
|
201 |
|
|
File outputFile = new File(outputDir,pdb.getPdbCode()+"_"+pdbChainCode+"_"+edgeType+"_"+cutoff+".graph"); |
202 |
|
|
graph.write_graph_to_file(outputFile.getAbsolutePath()); |
203 |
|
|
System.out.println("Wrote graph file "+outputFile.getAbsolutePath()+" from pdb file "+pdbfile); |
204 |
|
|
|
205 |
|
|
} catch (PdbfileFormatError e) { |
206 |
|
|
System.err.println("pdb file "+pdbfile+" doesn't have right format"); |
207 |
|
|
} catch (PdbChainCodeNotFoundError e) { |
208 |
|
|
System.err.println("chain code "+pdbChainCode+" wasn't found in file "+pdbfile); |
209 |
|
|
} |
210 |
|
|
} |
211 |
|
|
} |
212 |
|
|
|
213 |
|
|
} |