1 |
duarte |
123 |
package proteinstructure; |
2 |
|
|
|
3 |
|
|
import tools.MySQLConnection; |
4 |
duarte |
124 |
|
5 |
|
|
import java.security.AccessControlException; |
6 |
duarte |
123 |
import java.sql.Statement; |
7 |
|
|
import java.sql.ResultSet; |
8 |
|
|
import java.sql.SQLException; |
9 |
|
|
import java.util.ArrayList; |
10 |
|
|
import java.util.HashMap; |
11 |
|
|
import java.util.Collections; |
12 |
|
|
|
13 |
|
|
public class PdbaseInfo { |
14 |
|
|
public final static String MYSQLSERVER="white"; |
15 |
|
|
public final static String MYSQLUSER="duarte"; //TODO implement getuser method |
16 |
|
|
public final static String MYSQLPWD="nieve"; |
17 |
|
|
public final static String mymsdsdDB="my_msdsd_00_07_a"; |
18 |
|
|
public final static String msdsdDB="msdsd_00_07_a"; |
19 |
|
|
public final static String pdbaseDB="pdbase"; |
20 |
|
|
|
21 |
|
|
MySQLConnection conn; |
22 |
|
|
String accode=""; |
23 |
|
|
String chaincode=""; |
24 |
|
|
int model=1; |
25 |
|
|
int entrykey; |
26 |
|
|
String asymid; |
27 |
|
|
int entitykey; |
28 |
|
|
String alt_locs_sql_str; |
29 |
|
|
|
30 |
duarte |
124 |
PdbaseInfo (String accode, String chaincode, int model_serial, String db) throws PdbaseInconsistencyError, PdbaseAcCodeNotFoundError{ |
31 |
duarte |
123 |
this.accode=accode; |
32 |
|
|
this.chaincode=chaincode; |
33 |
|
|
this.model=model_serial; |
34 |
|
|
this.conn = new MySQLConnection(MYSQLSERVER,MYSQLUSER,MYSQLPWD,db); |
35 |
|
|
this.entrykey=get_entry_key(); |
36 |
|
|
this.asymid=get_asym_id(); |
37 |
|
|
this.entitykey=get_entity_key(); |
38 |
|
|
this.alt_locs_sql_str=get_atom_alt_locs(); |
39 |
|
|
|
40 |
|
|
} |
41 |
|
|
|
42 |
duarte |
124 |
PdbaseInfo (String accode, String chaincode, String db) throws PdbaseInconsistencyError, PdbaseAcCodeNotFoundError { |
43 |
duarte |
123 |
this(accode,chaincode,1,db); |
44 |
|
|
} |
45 |
duarte |
124 |
PdbaseInfo (String accode, String chaincode, int model_serial) throws PdbaseInconsistencyError, PdbaseAcCodeNotFoundError { |
46 |
duarte |
123 |
this(accode,chaincode,model_serial,pdbaseDB); |
47 |
|
|
} |
48 |
|
|
|
49 |
duarte |
124 |
PdbaseInfo (String accode, String chaincode) throws PdbaseInconsistencyError, PdbaseAcCodeNotFoundError { |
50 |
duarte |
123 |
this(accode,chaincode,1,pdbaseDB); |
51 |
|
|
} |
52 |
|
|
|
53 |
|
|
public void close() { |
54 |
|
|
conn.close(); |
55 |
|
|
} |
56 |
|
|
|
57 |
duarte |
124 |
public int get_entry_key() throws PdbaseAcCodeNotFoundError { |
58 |
duarte |
123 |
String sql="SELECT entry_key FROM struct WHERE entry_id='"+accode.toUpperCase()+"'"; |
59 |
|
|
try { |
60 |
|
|
Statement stmt = conn.createStatement(); |
61 |
|
|
ResultSet rsst = stmt.executeQuery(sql); |
62 |
|
|
if (rsst.next()) { |
63 |
|
|
entrykey = rsst.getInt(1); |
64 |
|
|
if (! rsst.isLast()) { |
65 |
|
|
System.err.println("More than 1 entry_key match for accession_code="+accode+", chain_pdb_code="+chaincode); |
66 |
duarte |
124 |
throw new PdbaseAcCodeNotFoundError(); |
67 |
duarte |
123 |
} |
68 |
|
|
} else { |
69 |
|
|
System.err.println("No entry_key match for accession_code="+accode+", chain_pdb_code="+chaincode); |
70 |
duarte |
124 |
throw new PdbaseAcCodeNotFoundError(); |
71 |
duarte |
123 |
} |
72 |
|
|
rsst.close(); |
73 |
|
|
stmt.close(); |
74 |
|
|
} catch (SQLException e) { |
75 |
|
|
e.printStackTrace(); |
76 |
|
|
} |
77 |
|
|
return entrykey; |
78 |
|
|
} |
79 |
|
|
|
80 |
duarte |
124 |
public String get_asym_id() throws PdbaseInconsistencyError { |
81 |
duarte |
123 |
String pdbstrandid=chaincode; |
82 |
|
|
if (chaincode.equals("NULL")){ |
83 |
|
|
pdbstrandid="A"; |
84 |
|
|
} |
85 |
|
|
String sql="SELECT asym_id " + |
86 |
|
|
" FROM pdbx_poly_seq_scheme " + |
87 |
|
|
" WHERE entry_key=" + entrykey + |
88 |
|
|
" AND pdb_strand_id='"+pdbstrandid+"' " + |
89 |
|
|
" LIMIT 1"; |
90 |
|
|
try { |
91 |
|
|
Statement stmt = conn.createStatement(); |
92 |
|
|
ResultSet rsst = stmt.executeQuery(sql); |
93 |
|
|
if (rsst.next()) { |
94 |
|
|
asymid = rsst.getString(1); |
95 |
|
|
} else { |
96 |
|
|
System.err.println("No asym_id match for entry_key="+entrykey+", pdb_strand_id="+chaincode); |
97 |
duarte |
124 |
throw new PdbaseInconsistencyError("No asym_id match for entry_key="+entrykey+", pdb_strand_id="+chaincode); |
98 |
duarte |
123 |
} |
99 |
|
|
rsst.close(); |
100 |
|
|
stmt.close(); |
101 |
|
|
} catch (SQLException e) { |
102 |
|
|
e.printStackTrace(); |
103 |
|
|
} |
104 |
|
|
return asymid; |
105 |
|
|
} |
106 |
|
|
|
107 |
duarte |
124 |
public int get_entity_key() throws PdbaseInconsistencyError { |
108 |
duarte |
123 |
String sql="SELECT entity_key " + |
109 |
|
|
" FROM struct_asym " + |
110 |
|
|
" WHERE entry_key="+ entrykey + |
111 |
|
|
" AND id='"+asymid+"'"; |
112 |
|
|
try { |
113 |
|
|
Statement stmt = conn.createStatement(); |
114 |
|
|
ResultSet rsst = stmt.executeQuery(sql); |
115 |
|
|
if (rsst.next()) { |
116 |
|
|
entitykey = rsst.getInt(1); |
117 |
|
|
if (! rsst.isLast()) { |
118 |
|
|
System.err.println("More than 1 entity_key match for entry_key="+entrykey+", asym_id="+asymid); |
119 |
duarte |
124 |
throw new PdbaseInconsistencyError("More than 1 entity_key match for entry_key="+entrykey+", asym_id="+asymid); |
120 |
duarte |
123 |
} |
121 |
|
|
} else { |
122 |
|
|
System.err.println("No entity_key match for entry_key="+entrykey+", asym_id="+asymid); |
123 |
duarte |
124 |
throw new PdbaseInconsistencyError("No entity_key match for entry_key="+entrykey+", asym_id="+asymid); |
124 |
duarte |
123 |
} |
125 |
|
|
rsst.close(); |
126 |
|
|
stmt.close(); |
127 |
|
|
} catch (SQLException e) { |
128 |
|
|
e.printStackTrace(); |
129 |
|
|
} |
130 |
|
|
return entitykey; |
131 |
|
|
} |
132 |
|
|
|
133 |
duarte |
124 |
public String get_atom_alt_locs() throws PdbaseInconsistencyError{ |
134 |
duarte |
123 |
ArrayList<String> alt_ids = new ArrayList<String>(); |
135 |
|
|
HashMap<String,Integer> alt_ids2keys = new HashMap<String,Integer>(); |
136 |
|
|
String alt_loc_field="label_alt_key"; |
137 |
|
|
String sql="SELECT id, atom_sites_alt_key FROM atom_sites_alt WHERE entry_key="+entrykey; |
138 |
|
|
try { |
139 |
|
|
Statement stmt = conn.createStatement(); |
140 |
|
|
ResultSet rsst = stmt.executeQuery(sql); |
141 |
|
|
int count=0; |
142 |
|
|
while (rsst.next()) { |
143 |
|
|
count++; |
144 |
|
|
alt_ids.add(rsst.getString(1)); |
145 |
|
|
alt_ids2keys.put(rsst.getString(1), rsst.getInt(2)); |
146 |
|
|
} |
147 |
|
|
if (count!=0){ |
148 |
|
|
if ((! alt_ids.contains(".")) || alt_ids.indexOf(".")!=alt_ids.lastIndexOf(".")){ // second term is a way of finding out if there is more than 1 ocurrence of "." in the ArrayList |
149 |
|
|
System.err.println("alt_codes exist for entry_key "+entrykey+" but there is either no default value '.' or more than 1 '.'. Something wrong with this entry_key or with "+pdbaseDB+" db!"); |
150 |
duarte |
124 |
throw new PdbaseInconsistencyError("alt_codes exist for entry_key "+entrykey+" but there is either no default value '.' or more than 1 '.'. Something wrong with this entry_key or with "+pdbaseDB+" db!"); |
151 |
duarte |
123 |
} |
152 |
|
|
alt_ids.remove("."); |
153 |
|
|
Collections.sort(alt_ids); |
154 |
|
|
String lowest_alt_id = alt_ids.get(0); |
155 |
|
|
alt_locs_sql_str = "("+alt_loc_field+"="+alt_ids2keys.get(".")+" OR "+alt_loc_field+"="+alt_ids2keys.get(lowest_alt_id)+")"; |
156 |
|
|
} else { |
157 |
|
|
alt_locs_sql_str=alt_loc_field+" IS NULL"; |
158 |
|
|
} |
159 |
|
|
|
160 |
|
|
rsst.close(); |
161 |
|
|
stmt.close(); |
162 |
|
|
} catch (SQLException e) { |
163 |
|
|
e.printStackTrace(); |
164 |
|
|
} |
165 |
|
|
|
166 |
|
|
return alt_locs_sql_str; |
167 |
|
|
} |
168 |
|
|
|
169 |
duarte |
124 |
public ArrayList<ArrayList> read_atomData() throws PdbaseInconsistencyError{ |
170 |
duarte |
123 |
ArrayList<ArrayList> resultset = new ArrayList<ArrayList>(); |
171 |
|
|
String sql = "SELECT id, label_atom_id, label_comp_id, label_asym_id, label_seq_id, Cartn_x, Cartn_y, Cartn_z " + |
172 |
|
|
" FROM atom_site " + |
173 |
|
|
" WHERE entry_key="+entrykey + |
174 |
|
|
" AND label_asym_id='"+asymid+"' " + |
175 |
|
|
" AND label_entity_key="+ entitykey + |
176 |
|
|
" AND model_num="+ model + |
177 |
|
|
" AND "+alt_locs_sql_str; |
178 |
|
|
try { |
179 |
|
|
Statement stmt = conn.createStatement(); |
180 |
|
|
ResultSet rsst = stmt.executeQuery(sql); |
181 |
duarte |
124 |
int count=0; |
182 |
duarte |
123 |
while (rsst.next()){ |
183 |
duarte |
124 |
count++; |
184 |
duarte |
123 |
ArrayList thisrecord = new ArrayList(); |
185 |
|
|
thisrecord.add(rsst.getInt(1)); //atomserial |
186 |
|
|
thisrecord.add(rsst.getString(2).trim()); // atom |
187 |
|
|
thisrecord.add(rsst.getString(3).trim()); // res_type |
188 |
|
|
thisrecord.add(rsst.getString(4).trim()); // chain |
189 |
|
|
thisrecord.add(rsst.getInt(5)); //res_serial |
190 |
|
|
thisrecord.add(rsst.getDouble(6)); // x |
191 |
|
|
thisrecord.add(rsst.getDouble(7)); // y |
192 |
|
|
thisrecord.add(rsst.getDouble(8)); // z |
193 |
|
|
|
194 |
|
|
resultset.add(thisrecord); |
195 |
|
|
} |
196 |
duarte |
124 |
if (count==0){ |
197 |
|
|
throw new PdbaseInconsistencyError("atom data query returned no data at all for entry_key="+entrykey+", asym_id="+asymid+", entity_key="+entitykey+", model_num="+model+", alt_locs_sql_str='"+alt_locs_sql_str+"'"); |
198 |
|
|
} |
199 |
duarte |
123 |
rsst.close(); |
200 |
|
|
stmt.close(); |
201 |
|
|
} catch (SQLException e) { |
202 |
|
|
e.printStackTrace(); |
203 |
|
|
} |
204 |
|
|
return resultset; |
205 |
|
|
} |
206 |
|
|
|
207 |
duarte |
124 |
public String read_seq() throws PdbaseInconsistencyError{ |
208 |
duarte |
123 |
String sequence=""; |
209 |
|
|
String pdbstrandid=chaincode; |
210 |
|
|
if (chaincode.equals("NULL")){ |
211 |
|
|
pdbstrandid="A"; |
212 |
|
|
} |
213 |
|
|
// we use seq_id+0 (implicitly converts to int) in ORDER BY because seq_id is varchar!! |
214 |
|
|
String sql="SELECT mon_id" + |
215 |
|
|
" FROM pdbx_poly_seq_scheme " + |
216 |
|
|
" WHERE entry_key=" + entrykey + |
217 |
|
|
" AND asym_id='"+asymid+"' " + |
218 |
|
|
" AND pdb_strand_id='"+pdbstrandid+"' " + |
219 |
|
|
" ORDER BY seq_id+0"; |
220 |
|
|
try { |
221 |
|
|
Statement stmt = conn.createStatement(); |
222 |
|
|
ResultSet rsst = stmt.executeQuery(sql); |
223 |
|
|
ArrayList<String> aalist=AA.aas(); |
224 |
|
|
int count=0; |
225 |
|
|
while (rsst.next()) { |
226 |
|
|
count++; |
227 |
|
|
String res_type = rsst.getString(1); |
228 |
|
|
if (aalist.contains(res_type)){ |
229 |
|
|
sequence+=AA.threeletter2oneletter(res_type); |
230 |
|
|
} else { |
231 |
|
|
sequence+="X"; |
232 |
|
|
} |
233 |
|
|
} |
234 |
|
|
if (count==0) { |
235 |
|
|
System.err.println("No sequence data match for entry_key="+entrykey+", asym_id="+asymid+", pdb_strand_id="+pdbstrandid); |
236 |
duarte |
124 |
throw new PdbaseInconsistencyError("No sequence data match for entry_key="+entrykey+", asym_id="+asymid+", pdb_strand_id="+pdbstrandid); |
237 |
duarte |
123 |
} |
238 |
|
|
rsst.close(); |
239 |
|
|
stmt.close(); |
240 |
|
|
} catch (SQLException e) { |
241 |
|
|
e.printStackTrace(); |
242 |
|
|
} |
243 |
|
|
|
244 |
|
|
return sequence; |
245 |
|
|
} |
246 |
|
|
} |