1 |
duarte |
207 |
package proteinstructure; |
2 |
|
|
|
3 |
|
|
import java.sql.ResultSet; |
4 |
|
|
import java.sql.SQLException; |
5 |
|
|
import java.sql.Statement; |
6 |
|
|
import java.util.ArrayList; |
7 |
|
|
import java.util.Collections; |
8 |
|
|
import java.util.HashMap; |
9 |
|
|
|
10 |
|
|
import tools.MySQLConnection; |
11 |
|
|
|
12 |
|
|
/** |
13 |
|
|
* A single chain pdb protein structure loaded from a PDBASE database |
14 |
|
|
* See http://openmms.sdsc.edu/OpenMMS-1.5.1_Std/openmms/docs/guides/PDBase.html to know what PDBASE is |
15 |
|
|
* |
16 |
|
|
* @author Jose Duarte |
17 |
|
|
* Class: PdbasePdb |
18 |
|
|
* Package: proteinstructure |
19 |
|
|
*/ |
20 |
|
|
public class PdbasePdb extends Pdb { |
21 |
|
|
|
22 |
|
|
private final static String MYSQLSERVER="white"; |
23 |
|
|
private final static String MYSQLUSER=MySQLConnection.getUserName(); |
24 |
|
|
private final static String MYSQLPWD="nieve"; |
25 |
|
|
private final static String DEFAULT_PDBASE_DB="pdbase"; |
26 |
|
|
|
27 |
|
|
private MySQLConnection conn; |
28 |
|
|
|
29 |
|
|
private int entrykey; |
30 |
|
|
private String asymid; |
31 |
|
|
private int entitykey; |
32 |
|
|
private String alt_locs_sql_str; |
33 |
|
|
|
34 |
|
|
/** |
35 |
|
|
* Constructs Pdb object given pdb code and pdb chain code. |
36 |
|
|
* Model will be DEFAULT_MODEL |
37 |
|
|
* MySQLConnection is taken from defaults in PdbasePdb class: MYSQLSERVER, MYSQLUSER, MYSQLPWD |
38 |
|
|
* Database is taken from default pdbase database in PdbasePdb class: DEFAULT_PDBASE_DB |
39 |
|
|
* @param pdbCode |
40 |
|
|
* @param pdbChainCode |
41 |
|
|
* @throws PdbaseInconsistencyError |
42 |
|
|
* @throws PdbaseAcCodeNotFoundError |
43 |
|
|
* @throws SQLException |
44 |
|
|
*/ |
45 |
|
|
public PdbasePdb (String pdbCode, String pdbChainCode) throws PdbaseInconsistencyError, PdbaseAcCodeNotFoundError, SQLException { |
46 |
|
|
this(pdbCode, pdbChainCode, DEFAULT_MODEL, DEFAULT_PDBASE_DB, new MySQLConnection(MYSQLSERVER,MYSQLUSER,MYSQLPWD)); |
47 |
|
|
} |
48 |
|
|
|
49 |
|
|
/** |
50 |
|
|
* Constructs Pdb object given pdb code, pdb chain code, source db and a MySQLConnection |
51 |
|
|
* Model will be DEFAULT_MODEL |
52 |
|
|
* The db must be a pdbase database |
53 |
|
|
* @param pdbCode |
54 |
|
|
* @param pdbChainCode |
55 |
|
|
* @param db |
56 |
|
|
* @param conn |
57 |
|
|
* @throws PdbaseInconsistencyError |
58 |
|
|
* @throws PdbaseAcCodeNotFoundError |
59 |
duarte |
210 |
* @throws SQLException |
60 |
duarte |
207 |
*/ |
61 |
duarte |
210 |
public PdbasePdb (String pdbCode, String pdbChainCode, String db, MySQLConnection conn) throws PdbaseInconsistencyError, PdbaseAcCodeNotFoundError, SQLException { |
62 |
duarte |
207 |
this(pdbCode,pdbChainCode,DEFAULT_MODEL,db, conn); |
63 |
|
|
} |
64 |
|
|
|
65 |
|
|
/** |
66 |
|
|
* Constructs Pdb object given pdb code, pdb chain code and model serial. |
67 |
|
|
* MySQLConnection is taken from defaults in PdbasePdb class: MYSQLSERVER, MYSQLUSER, MYSQLPWD |
68 |
|
|
* Database is taken from default pdbase database in PdbasePdb class: DEFAULT_PDBASE_DB |
69 |
|
|
* @param pdbCode |
70 |
|
|
* @param pdbChainCode |
71 |
|
|
* @param model_serial |
72 |
|
|
* @throws PdbaseInconsistencyError |
73 |
|
|
* @throws PdbaseAcCodeNotFoundError |
74 |
|
|
* @throws SQLException |
75 |
|
|
*/ |
76 |
|
|
public PdbasePdb (String pdbCode, String pdbChainCode, int model_serial) throws PdbaseInconsistencyError, PdbaseAcCodeNotFoundError, SQLException { |
77 |
|
|
this(pdbCode, pdbChainCode, model_serial, DEFAULT_PDBASE_DB, new MySQLConnection(MYSQLSERVER,MYSQLUSER,MYSQLPWD)); |
78 |
|
|
} |
79 |
|
|
|
80 |
|
|
/** |
81 |
|
|
* Constructs Pdb object given pdb code, pdb chain code, model serial, source db and a MySQLConnection. |
82 |
|
|
* The db must be a pdbase database |
83 |
|
|
* @param pdbCode |
84 |
|
|
* @param pdbChainCode |
85 |
|
|
* @param model_serial |
86 |
|
|
* @param db |
87 |
|
|
* @param conn |
88 |
|
|
* @throws PdbaseInconsistencyError |
89 |
|
|
* @throws PdbaseAcCodeNotFoundError |
90 |
duarte |
210 |
* @throws SQLException |
91 |
duarte |
207 |
*/ |
92 |
duarte |
210 |
public PdbasePdb (String pdbCode, String pdbChainCode, int model_serial, String db, MySQLConnection conn) throws PdbaseInconsistencyError, PdbaseAcCodeNotFoundError, SQLException { |
93 |
duarte |
207 |
this.pdbCode=pdbCode; |
94 |
|
|
this.pdbChainCode=pdbChainCode; |
95 |
|
|
this.model=model_serial; |
96 |
|
|
this.db=db; |
97 |
|
|
|
98 |
|
|
this.conn = conn; |
99 |
|
|
this.entrykey=get_entry_key(); |
100 |
|
|
this.asymid=get_asym_id(); // sets asymid and chainCode |
101 |
|
|
this.entitykey=get_entity_key(); |
102 |
|
|
this.alt_locs_sql_str=get_atom_alt_locs(); |
103 |
|
|
|
104 |
|
|
this.chainCode = getChainCode(); |
105 |
|
|
this.sequence = read_seq(); |
106 |
|
|
this.pdbresser2resser = get_ressers_mapping(); |
107 |
|
|
|
108 |
|
|
this.read_atomData(); // populates resser_atom2atomserial, resser2restype, atomser2coord, atomser2resser |
109 |
|
|
|
110 |
|
|
// we initialise resser2pdbresser from the pdbresser2resser HashMap |
111 |
|
|
this.resser2pdbresser = new HashMap<Integer, String>(); |
112 |
|
|
for (String pdbresser:pdbresser2resser.keySet()){ |
113 |
|
|
resser2pdbresser.put(pdbresser2resser.get(pdbresser), pdbresser); |
114 |
|
|
} |
115 |
|
|
} |
116 |
|
|
|
117 |
duarte |
210 |
private int get_entry_key() throws PdbaseAcCodeNotFoundError, SQLException { |
118 |
duarte |
207 |
String sql="SELECT entry_key FROM "+db+".struct WHERE entry_id='"+pdbCode.toUpperCase()+"'"; |
119 |
duarte |
210 |
Statement stmt = conn.createStatement(); |
120 |
|
|
ResultSet rsst = stmt.executeQuery(sql); |
121 |
|
|
if (rsst.next()) { |
122 |
|
|
entrykey = rsst.getInt(1); |
123 |
|
|
if (! rsst.isLast()) { |
124 |
|
|
System.err.println("More than 1 entry_key match for accession_code="+pdbCode+", chain_pdb_code="+pdbChainCode); |
125 |
|
|
throw new PdbaseAcCodeNotFoundError(); |
126 |
duarte |
207 |
} |
127 |
duarte |
210 |
} else { |
128 |
|
|
System.err.println("No entry_key match for accession_code="+pdbCode+", chain_pdb_code="+pdbChainCode); |
129 |
|
|
throw new PdbaseAcCodeNotFoundError(); |
130 |
duarte |
207 |
} |
131 |
duarte |
210 |
rsst.close(); |
132 |
|
|
stmt.close(); |
133 |
duarte |
207 |
return entrykey; |
134 |
|
|
} |
135 |
|
|
|
136 |
duarte |
210 |
private String get_asym_id() throws PdbaseInconsistencyError, SQLException { |
137 |
duarte |
207 |
String pdbstrandid=pdbChainCode; |
138 |
|
|
if (pdbChainCode.equals("NULL")){ |
139 |
|
|
pdbstrandid="A"; |
140 |
|
|
} |
141 |
|
|
String sql="SELECT asym_id " + |
142 |
|
|
" FROM "+db+".pdbx_poly_seq_scheme " + |
143 |
|
|
" WHERE entry_key=" + entrykey + |
144 |
|
|
" AND pdb_strand_id='"+pdbstrandid+"' " + |
145 |
|
|
" LIMIT 1"; |
146 |
duarte |
210 |
|
147 |
|
|
Statement stmt = conn.createStatement(); |
148 |
|
|
ResultSet rsst = stmt.executeQuery(sql); |
149 |
|
|
if (rsst.next()) { |
150 |
|
|
asymid = rsst.getString(1); |
151 |
|
|
} else { |
152 |
|
|
System.err.println("No asym_id match for entry_key="+entrykey+", pdb_strand_id="+pdbChainCode); |
153 |
|
|
throw new PdbaseInconsistencyError("No asym_id match for entry_key="+entrykey+", pdb_strand_id="+pdbChainCode); |
154 |
duarte |
207 |
} |
155 |
duarte |
210 |
rsst.close(); |
156 |
|
|
stmt.close(); |
157 |
duarte |
207 |
// we set the internal chain identifier chainCode from asymid |
158 |
|
|
chainCode = asymid; |
159 |
|
|
return asymid; |
160 |
|
|
} |
161 |
|
|
|
162 |
duarte |
210 |
private int get_entity_key() throws PdbaseInconsistencyError, SQLException { |
163 |
duarte |
207 |
String sql="SELECT entity_key " + |
164 |
|
|
" FROM "+db+".struct_asym " + |
165 |
|
|
" WHERE entry_key="+ entrykey + |
166 |
|
|
" AND id='"+asymid+"'"; |
167 |
duarte |
210 |
|
168 |
|
|
Statement stmt = conn.createStatement(); |
169 |
|
|
ResultSet rsst = stmt.executeQuery(sql); |
170 |
|
|
if (rsst.next()) { |
171 |
|
|
entitykey = rsst.getInt(1); |
172 |
|
|
if (! rsst.isLast()) { |
173 |
|
|
System.err.println("More than 1 entity_key match for entry_key="+entrykey+", asym_id="+asymid); |
174 |
|
|
throw new PdbaseInconsistencyError("More than 1 entity_key match for entry_key="+entrykey+", asym_id="+asymid); |
175 |
duarte |
207 |
} |
176 |
duarte |
210 |
} else { |
177 |
|
|
System.err.println("No entity_key match for entry_key="+entrykey+", asym_id="+asymid); |
178 |
|
|
throw new PdbaseInconsistencyError("No entity_key match for entry_key="+entrykey+", asym_id="+asymid); |
179 |
duarte |
207 |
} |
180 |
duarte |
210 |
rsst.close(); |
181 |
|
|
stmt.close(); |
182 |
duarte |
207 |
return entitykey; |
183 |
|
|
} |
184 |
|
|
|
185 |
duarte |
210 |
private String get_atom_alt_locs() throws PdbaseInconsistencyError, SQLException{ |
186 |
duarte |
207 |
ArrayList<String> alt_ids = new ArrayList<String>(); |
187 |
|
|
HashMap<String,Integer> alt_ids2keys = new HashMap<String,Integer>(); |
188 |
|
|
String alt_loc_field="label_alt_key"; |
189 |
|
|
String sql="SELECT id, atom_sites_alt_key FROM "+db+".atom_sites_alt WHERE entry_key="+entrykey; |
190 |
|
|
|
191 |
duarte |
210 |
Statement stmt = conn.createStatement(); |
192 |
|
|
ResultSet rsst = stmt.executeQuery(sql); |
193 |
|
|
int count=0; |
194 |
|
|
while (rsst.next()) { |
195 |
|
|
count++; |
196 |
|
|
alt_ids.add(rsst.getString(1)); |
197 |
|
|
alt_ids2keys.put(rsst.getString(1), rsst.getInt(2)); |
198 |
duarte |
207 |
} |
199 |
duarte |
210 |
if (count!=0){ |
200 |
|
|
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 |
201 |
|
|
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 "+DEFAULT_PDBASE_DB+" db!"); |
202 |
|
|
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 "+DEFAULT_PDBASE_DB+" db!"); |
203 |
|
|
} |
204 |
|
|
alt_ids.remove("."); |
205 |
|
|
Collections.sort(alt_ids); |
206 |
|
|
String lowest_alt_id = alt_ids.get(0); |
207 |
|
|
alt_locs_sql_str = "("+alt_loc_field+"="+alt_ids2keys.get(".")+" OR "+alt_loc_field+"="+alt_ids2keys.get(lowest_alt_id)+")"; |
208 |
|
|
} else { |
209 |
|
|
alt_locs_sql_str=alt_loc_field+" IS NULL"; |
210 |
|
|
} |
211 |
|
|
|
212 |
|
|
rsst.close(); |
213 |
|
|
stmt.close(); |
214 |
|
|
|
215 |
duarte |
207 |
return alt_locs_sql_str; |
216 |
|
|
} |
217 |
|
|
|
218 |
duarte |
210 |
private void read_atomData() throws PdbaseInconsistencyError, SQLException{ |
219 |
duarte |
207 |
resser_atom2atomserial = new HashMap<String,Integer>(); |
220 |
|
|
resser2restype = new HashMap<Integer,String>(); |
221 |
|
|
atomser2coord = new HashMap<Integer,Double[]>(); |
222 |
|
|
atomser2resser = new HashMap<Integer,Integer>(); |
223 |
|
|
|
224 |
|
|
|
225 |
|
|
String sql = "SELECT id, label_atom_id, label_comp_id, label_seq_id, Cartn_x, Cartn_y, Cartn_z " + |
226 |
|
|
" FROM "+db+".atom_site " + |
227 |
|
|
" WHERE entry_key="+entrykey + |
228 |
|
|
" AND label_asym_id='"+asymid+"' " + |
229 |
|
|
" AND label_entity_key="+ entitykey + |
230 |
|
|
" AND model_num="+ model + |
231 |
|
|
" AND "+alt_locs_sql_str; |
232 |
|
|
|
233 |
duarte |
210 |
Statement stmt = conn.createStatement(); |
234 |
|
|
ResultSet rsst = stmt.executeQuery(sql); |
235 |
|
|
int count=0; |
236 |
|
|
while (rsst.next()){ |
237 |
|
|
count++; |
238 |
|
|
|
239 |
|
|
int atomserial = rsst.getInt(1); // atomserial |
240 |
|
|
String atom = rsst.getString(2).trim(); // atom |
241 |
|
|
String res_type = rsst.getString(3).trim(); // res_type |
242 |
|
|
int res_serial = rsst.getInt(4); // res_serial |
243 |
|
|
double x = rsst.getDouble(5); // x |
244 |
|
|
double y = rsst.getDouble(6); // y |
245 |
|
|
double z = rsst.getDouble(7); // z |
246 |
|
|
Double[] coords = {x, y, z}; |
247 |
|
|
ArrayList<String> aalist=AA.aas(); |
248 |
|
|
if (aalist.contains(res_type)) { |
249 |
|
|
atomser2coord.put(atomserial, coords); |
250 |
|
|
atomser2resser.put(atomserial, res_serial); |
251 |
|
|
resser2restype.put(res_serial, res_type); |
252 |
|
|
ArrayList<String> atomlist = aas2atoms.get(res_type); |
253 |
|
|
if (atomlist.contains(atom)){ |
254 |
|
|
resser_atom2atomserial.put(res_serial+"_"+atom, atomserial); |
255 |
duarte |
207 |
} |
256 |
|
|
} |
257 |
duarte |
210 |
|
258 |
duarte |
207 |
} |
259 |
duarte |
210 |
if (count==0){ |
260 |
|
|
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+"'"); |
261 |
|
|
} |
262 |
|
|
rsst.close(); |
263 |
|
|
stmt.close(); |
264 |
duarte |
207 |
} |
265 |
|
|
|
266 |
duarte |
210 |
private String read_seq() throws PdbaseInconsistencyError, SQLException{ |
267 |
duarte |
207 |
String sequence=""; |
268 |
|
|
String pdbstrandid=pdbChainCode; |
269 |
|
|
if (pdbChainCode.equals("NULL")){ |
270 |
|
|
pdbstrandid="A"; |
271 |
|
|
} |
272 |
|
|
// we use seq_id+0 (implicitly converts to int) in ORDER BY because seq_id is varchar!! |
273 |
|
|
String sql="SELECT mon_id" + |
274 |
|
|
" FROM "+db+".pdbx_poly_seq_scheme " + |
275 |
|
|
" WHERE entry_key=" + entrykey + |
276 |
|
|
" AND asym_id='"+asymid+"' " + |
277 |
|
|
" AND pdb_strand_id='"+pdbstrandid+"' " + |
278 |
|
|
" ORDER BY seq_id+0"; |
279 |
|
|
|
280 |
duarte |
210 |
Statement stmt = conn.createStatement(); |
281 |
|
|
ResultSet rsst = stmt.executeQuery(sql); |
282 |
|
|
ArrayList<String> aalist=AA.aas(); |
283 |
|
|
int count=0; |
284 |
|
|
while (rsst.next()) { |
285 |
|
|
count++; |
286 |
|
|
String res_type = rsst.getString(1); |
287 |
|
|
if (aalist.contains(res_type)){ |
288 |
|
|
sequence+=AA.threeletter2oneletter(res_type); |
289 |
|
|
} else { |
290 |
|
|
sequence+=NONSTANDARD_AA_LETTER; |
291 |
|
|
} |
292 |
|
|
} |
293 |
|
|
if (count==0) { |
294 |
|
|
System.err.println("No sequence data match for entry_key="+entrykey+", asym_id="+asymid+", pdb_strand_id="+pdbstrandid); |
295 |
|
|
throw new PdbaseInconsistencyError("No sequence data match for entry_key="+entrykey+", asym_id="+asymid+", pdb_strand_id="+pdbstrandid); |
296 |
|
|
} |
297 |
|
|
rsst.close(); |
298 |
|
|
stmt.close(); |
299 |
|
|
|
300 |
duarte |
207 |
return sequence; |
301 |
|
|
} |
302 |
|
|
|
303 |
duarte |
210 |
private HashMap<String,Integer> get_ressers_mapping() throws PdbaseInconsistencyError, SQLException{ |
304 |
duarte |
207 |
String pdbstrandid=pdbChainCode; |
305 |
|
|
if (pdbChainCode.equals("NULL")){ |
306 |
|
|
pdbstrandid="A"; |
307 |
|
|
} |
308 |
|
|
|
309 |
|
|
HashMap<String,Integer> map = new HashMap<String, Integer>(); |
310 |
|
|
String sql="SELECT seq_id, concat(auth_seq_num,IF(pdb_ins_code='.','',pdb_ins_code))" + |
311 |
|
|
" FROM "+db+".pdbx_poly_seq_scheme " + |
312 |
|
|
" WHERE entry_key=" + entrykey + |
313 |
|
|
" AND asym_id='"+asymid+"' " + |
314 |
|
|
" AND pdb_strand_id='"+pdbstrandid+"' " + |
315 |
|
|
" AND auth_seq_num!='?'" + |
316 |
|
|
" ORDER BY seq_id+0"; |
317 |
duarte |
210 |
|
318 |
|
|
Statement stmt = conn.createStatement(); |
319 |
|
|
ResultSet rsst = stmt.executeQuery(sql); |
320 |
|
|
int count=0; |
321 |
|
|
while (rsst.next()) { |
322 |
|
|
count++; |
323 |
|
|
int resser = Integer.parseInt(rsst.getString(1)); |
324 |
|
|
String pdbresser = rsst.getString(2); |
325 |
|
|
map.put(pdbresser, resser); |
326 |
|
|
} |
327 |
|
|
if (count==0) { |
328 |
|
|
System.err.println("No residue serials mapping data match for entry_key="+entrykey+", asym_id="+asymid+", pdb_strand_id="+pdbstrandid); |
329 |
|
|
throw new PdbaseInconsistencyError("No residue serials mapping data match for entry_key="+entrykey+", asym_id="+asymid+", pdb_strand_id="+pdbstrandid); |
330 |
duarte |
207 |
} |
331 |
duarte |
210 |
rsst.close(); |
332 |
|
|
stmt.close(); |
333 |
duarte |
207 |
|
334 |
|
|
return map; |
335 |
|
|
} |
336 |
|
|
|
337 |
|
|
} |