Skip to content

Commit 97924c5

Browse files
sblivensbliven
committed
First round of fixes to ECOD format
Format was updated slightly in response to our testing, but some problems and inconsistencies remain.
1 parent 9cf551a commit 97924c5

File tree

3 files changed

+134
-59
lines changed

3 files changed

+134
-59
lines changed

‎biojava-integrationtest/src/test/java/org/biojava/nbio/structure/test/domain/EcodInstallationTest.java‎

Lines changed: 26 additions & 6 deletions
Original file line numberDiff line numberDiff line change
@@ -27,7 +27,6 @@
2727
import java.util.ArrayList;
2828
import java.util.Arrays;
2929
import java.util.Collections;
30-
import java.util.HashMap;
3130
import java.util.HashSet;
3231
import java.util.List;
3332
import java.util.Set;
@@ -55,7 +54,7 @@
5554
public class EcodInstallationTest {
5655

5756
private static final Logger logger = LoggerFactory.getLogger(EcodInstallationTest.class);
58-
private static final String VERSION = "develop77";
57+
private static final String VERSION = "develop78";
5958

6059
static {
6160
//System.setProperty("Log4jContextSelector", "org.apache.logging.log4j.core.async.AsyncLoggerContextSelector");
@@ -80,10 +79,13 @@ public void testDownloads() throws IOException {
8079

8180
@Test
8281
public void testAllDomains() throws IOException {
82+
int expected;
8383
EcodDatabase ecod = EcodFactory.getEcodDatabase(VERSION);
8484

8585
List<EcodDomain> domains = ecod.getAllDomains();
86-
assertEquals("Wrong number of domains",423779,domains.size());
86+
expected = 423779; //version77
87+
expected = 423869; //version78
88+
assertEquals("Wrong number of domains",expected,domains.size());
8789
}
8890

8991
@Test
@@ -122,14 +124,15 @@ public void testParsing() throws IOException {
122124
expected = new EcodDomain(
123125
// Long uid, String domainId, Boolean manual,
124126
20669l, "e1lyw.1", null,
125-
// Integer xGroup, Integer hGroup, Integer tGroup, String pdbId,
126-
1,1,1,"1lyw",
127+
// Integer xGroup, Integer hGroup, Integer tGroup, Integer fGroup, String pdbId,
128+
1,1,1,2,"1lyw",
127129
// String chainId, String range, String architectureName,
128130
".", "A:3-97,B:106-346", "beta barrels",
129131
// String xGroupName, String hGroupName, String tGroupName,
130132
// String fGroupName, Boolean isAssembly, List<String> ligands
131133
"cradle loop barrel", "RIFT-related", "acid protease",
132-
"UNK_F_TYPE", false, Collections.singleton("EPE")
134+
"EF00710",//"UNK_F_TYPE",
135+
20669l, Collections.singleton("EPE")
133136
);
134137
assertEquals(ecodId,expected,domain);
135138

@@ -221,4 +224,21 @@ public void testVersion() throws IOException {
221224
assertNotNull(version);
222225
assertNotEquals("latest", version);
223226
}
227+
228+
@Test
229+
public void testAllVersions() throws IOException {
230+
int firstVersion = 45;
231+
int lastVersion = 78;
232+
List<String> versions = new ArrayList<String>();
233+
234+
for(int version = firstVersion; version<= lastVersion;version++) {
235+
versions.add("develop"+version);
236+
}
237+
versions.add("latest");
238+
for(String version : versions) {
239+
EcodInstallation ecod = (EcodInstallation)EcodFactory.getEcodDatabase(version);
240+
ecod.getAllDomains();
241+
System.out.println(version +" -> "+ ecod.getVersion());
242+
}
243+
}
224244
}

‎biojava-structure/src/main/java/org/biojava/nbio/structure/ecod/EcodDomain.java‎

Lines changed: 49 additions & 34 deletions
Original file line numberDiff line numberDiff line change
@@ -21,7 +21,6 @@
2121

2222
import java.io.Serializable;
2323
import java.util.HashSet;
24-
import java.util.List;
2524
import java.util.Set;
2625

2726
/**
@@ -34,7 +33,7 @@ public class EcodDomain implements Serializable, Cloneable {
3433
Column 1: ECOD uid - internal domain unique identifier
3534
Column 2: ECOD domain id - domain identifier
3635
Column 3: ECOD representative status - manual (curated) or automated nonrep
37-
Column 4: ECOD hierachy identifier - [X-group].[H-group].{T-group]
36+
Column 4: ECOD hierachy identifier - [X-group].[H-group].{T-group].[F-group]
3837
Column 5: PDB identifier
3938
Column 6: Chain identifier (note: case-sensitive)
4039
Column 7: PDB residue number range
@@ -54,15 +53,14 @@ public class EcodDomain implements Serializable, Cloneable {
5453

5554
private static final long serialVersionUID = -7760082165560332048L;
5655

57-
/** String for unclassified F-groups */
58-
public static final String F_UNCLASSFIED = "F_UNCLASSIFIED";
5956

6057
private Long uid;
6158
private String domainId;
6259
private Boolean manual;
6360
private Integer xGroup;
6461
private Integer hGroup;
6562
private Integer tGroup;
63+
private Integer fGroup;
6664
private String pdbId;
6765
private String chainId;
6866
private String range;
@@ -71,23 +69,24 @@ public class EcodDomain implements Serializable, Cloneable {
7169
private String hGroupName;
7270
private String tGroupName;
7371
private String fGroupName;
74-
private Boolean isAssembly; // Maybe should be a list, according to description?
72+
private Long assemblyId; //for non-assemblies, matches the uid.
7573
private Set<String> ligands;
7674

7775
/** Default constructor with all null properties */
7876
public EcodDomain() {}
7977

8078
public EcodDomain(Long uid, String domainId, Boolean manual,
81-
Integer xGroup, Integer hGroup, Integer tGroup, String pdbId,
79+
Integer xGroup, Integer hGroup, Integer tGroup, Integer fGroup, String pdbId,
8280
String chainId, String range, String architectureName,
8381
String xGroupName, String hGroupName, String tGroupName,
84-
String fGroupName, Boolean isAssembly, Set<String> ligands) {
82+
String fGroupName, Long assemblyId, Set<String> ligands) {
8583
this.uid = uid;
8684
this.domainId = domainId;
8785
this.manual = manual;
8886
this.xGroup = xGroup;
8987
this.hGroup = hGroup;
9088
this.tGroup = tGroup;
89+
this.fGroup = fGroup;
9190
this.pdbId = pdbId;
9291
this.chainId = chainId;
9392
this.range = range;
@@ -96,7 +95,7 @@ public EcodDomain(Long uid, String domainId, Boolean manual,
9695
this.hGroupName = hGroupName;
9796
this.tGroupName = tGroupName;
9897
this.fGroupName = fGroupName;
99-
this.isAssembly = isAssembly;
98+
this.assemblyId = assemblyId;
10099
this.ligands = ligands;
101100
}
102101
public EcodDomain(String domainId) {
@@ -109,6 +108,7 @@ public EcodDomain(EcodDomain o) {
109108
this.xGroup = o.xGroup;
110109
this.hGroup = o.hGroup;
111110
this.tGroup = o.tGroup;
111+
this.fGroup = o.fGroup;
112112
this.pdbId = o.pdbId;
113113
this.chainId = o.chainId;
114114
this.range = o.range;
@@ -117,7 +117,7 @@ public EcodDomain(EcodDomain o) {
117117
this.hGroupName = o.hGroupName;
118118
this.tGroupName = o.tGroupName;
119119
this.fGroupName = o.fGroupName;
120-
this.isAssembly = o.isAssembly;
120+
this.assemblyId = o.assemblyId;
121121
this.ligands = new HashSet<String>(o.ligands);
122122
}
123123

@@ -144,24 +144,30 @@ public Boolean getManual() {
144144
public void setManual(Boolean manual) {
145145
this.manual = manual;
146146
}
147-
public Integer getxGroup() {
147+
public Integer getXGroup() {
148148
return xGroup;
149149
}
150-
public void setxGroup(Integer xGroup) {
150+
public void setXGroup(Integer xGroup) {
151151
this.xGroup = xGroup;
152152
}
153-
public Integer gethGroup() {
153+
public Integer getHGroup() {
154154
return hGroup;
155155
}
156-
public void sethGroup(Integer hGroup) {
156+
public void setHGroup(Integer hGroup) {
157157
this.hGroup = hGroup;
158158
}
159-
public Integer gettGroup() {
159+
public Integer getTGroup() {
160160
return tGroup;
161161
}
162-
public void settGroup(Integer tGroup) {
162+
public void setTGroup(Integer tGroup) {
163163
this.tGroup = tGroup;
164164
}
165+
public Integer getFGroup() {
166+
return fGroup;
167+
}
168+
public void setFGroup(Integer fGroup) {
169+
this.fGroup = fGroup;
170+
}
165171
public String getPdbId() {
166172
return pdbId;
167173
}
@@ -186,35 +192,38 @@ public String getArchitectureName() {
186192
public void setArchitectureName(String architectureName) {
187193
this.architectureName = architectureName;
188194
}
189-
public String getxGroupName() {
195+
public String getXGroupName() {
190196
return xGroupName;
191197
}
192-
public void setxGroupName(String xGroupName) {
198+
public void setXGroupName(String xGroupName) {
193199
this.xGroupName = xGroupName;
194200
}
195-
public String gethGroupName() {
201+
public String getHGroupName() {
196202
return hGroupName;
197203
}
198-
public void sethGroupName(String hGroupName) {
204+
public void setHGroupName(String hGroupName) {
199205
this.hGroupName = hGroupName;
200206
}
201-
public String gettGroupName() {
207+
public String getTGroupName() {
202208
return tGroupName;
203209
}
204-
public void settGroupName(String tGroupName) {
210+
public void setGroupName(String tGroupName) {
205211
this.tGroupName = tGroupName;
206212
}
207-
public String getfGroupName() {
213+
public String getFGroupName() {
208214
return fGroupName;
209215
}
210-
public void setfGroupName(String fGroupName) {
216+
public void setFGroupName(String fGroupName) {
211217
this.fGroupName = fGroupName;
212218
}
213-
public Boolean getIsAssembly() {
214-
return isAssembly;
219+
/**
220+
* @return The assembly ID, or the DomainId if not in an assembly, or null if unknown.
221+
*/
222+
public Long getAssemblyId() {
223+
return assemblyId;
215224
}
216-
public void setIsAssembly(Boolean isAssembly) {
217-
this.isAssembly = isAssembly;
225+
public void setAssemblyId(Long assemblyId) {
226+
this.assemblyId = assemblyId;
218227
}
219228
public Set<String> getLigands() {
220229
return ligands;
@@ -230,12 +239,12 @@ public void setLigands(Set<String> ligands) {
230239
public String toString() {
231240
return "EcodDomain [uid=" + uid + ", domainId=" + domainId
232241
+ ", manual=" + manual + ", xGroup=" + xGroup + ", hGroup="
233-
+ hGroup + ", tGroup=" + tGroup + ", pdbId=" + pdbId
242+
+ hGroup + ", tGroup=" + tGroup + ", fGroup="+ fGroup + ", pdbId=" + pdbId
234243
+ ", chainId=" + chainId + ", range=" + range
235244
+ ", architectureName=" + architectureName + ", xGroupName="
236245
+ xGroupName + ", hGroupName=" + hGroupName + ", tGroupName="
237-
+ tGroupName + ", fGroupName=" + fGroupName + ", isAssembly="
238-
+ isAssembly + ", ligands=" + ligands + "]";
246+
+ tGroupName + ", fGroupName=" + fGroupName + ", assemblyId="
247+
+ assemblyId + ", ligands=" + ligands + "]";
239248
}
240249

241250
/* (non-Javadoc)
@@ -253,11 +262,12 @@ public int hashCode() {
253262
+ ((domainId == null) ? 0 : domainId.hashCode());
254263
result = prime * result
255264
+ ((fGroupName == null) ? 0 : fGroupName.hashCode());
265+
result = prime * result + ((fGroup == null) ? 0 : fGroup.hashCode());
256266
result = prime * result + ((hGroup == null) ? 0 : hGroup.hashCode());
257267
result = prime * result
258268
+ ((hGroupName == null) ? 0 : hGroupName.hashCode());
259269
result = prime * result
260-
+ ((isAssembly == null) ? 0 : isAssembly.hashCode());
270+
+ ((assemblyId == null) ? 0 : assemblyId.hashCode());
261271
result = prime * result + ((ligands == null) ? 0 : ligands.hashCode());
262272
result = prime * result + ((manual == null) ? 0 : manual.hashCode());
263273
result = prime * result + ((pdbId == null) ? 0 : pdbId.hashCode());
@@ -304,6 +314,11 @@ public boolean equals(Object obj) {
304314
return false;
305315
} else if (!fGroupName.equals(other.fGroupName))
306316
return false;
317+
if (fGroup == null) {
318+
if (other.fGroup != null)
319+
return false;
320+
} else if (!fGroup.equals(other.fGroup))
321+
return false;
307322
if (hGroup == null) {
308323
if (other.hGroup != null)
309324
return false;
@@ -314,10 +329,10 @@ public boolean equals(Object obj) {
314329
return false;
315330
} else if (!hGroupName.equals(other.hGroupName))
316331
return false;
317-
if (isAssembly == null) {
318-
if (other.isAssembly != null)
332+
if (assemblyId == null) {
333+
if (other.assemblyId != null)
319334
return false;
320-
} else if (!isAssembly.equals(other.isAssembly))
335+
} else if (!assemblyId.equals(other.assemblyId))
321336
return false;
322337
if (ligands == null) {
323338
if (other.ligands != null)

0 commit comments

Comments
 (0)