Skip to content

Commit 4e1e063

Browse files
IgorRodchenkovIgorRodchenkov
committed
Fixed ChEBI Converter (recent chebi.obo differs from that in the past in e.g. how SMILES and InChI values are there defined).
Added/updated JSONLD converter related tests (paxtools-jsonld is back to jena v3 from v4/v5 which generated quite different output/errors). Migrated from weird jakarta.json.Json* to jackson in BasicController.
1 parent 290e622 commit 4e1e063

File tree

10 files changed

+618
-61
lines changed

10 files changed

+618
-61
lines changed

‎src/main/java/cpath/converter/ChebiOboConverter.java

Lines changed: 47 additions & 7 deletions
Original file line numberDiff line numberDiff line change
@@ -42,8 +42,9 @@ class ChebiOboConverter extends BaseConverter
4242
private final String _NAME = "name: ";
4343
private final String _DEF = "def: ";
4444
private final String _SYNONYM = "synonym: ";
45+
private final String _PROPERTY_VALUE = "property_value: "; //new fields in recent chebi.obo
4546

46-
//to extract a text value between quotation marks from 'def:' and 'synonym:' lines:
47+
//to extract a text value between quotation marks in the 'def:', 'synonym:', 'property_value:' lines:
4748
private final Pattern namePattern = Pattern.compile("\"(.+?)\"");
4849
//to extract ID, DB values from 'xref:' lines:
4950
//(since ChEBI OBO format has been slightly changed in 2017, pattern was updated)
@@ -95,6 +96,9 @@ else if (line.startsWith(_ALT_ID)) {
9596
else if (line.startsWith(_SYNONYM)) {
9697
updateMapEntry(chebiEntryMap, _SYNONYM, line);
9798
}
99+
else if (line.startsWith(_PROPERTY_VALUE)) {
100+
updateMapEntry(chebiEntryMap, _PROPERTY_VALUE, line);
101+
}
98102
else if (line.startsWith(_XREF)) {
99103
updateMapEntry(chebiEntryMap, _XREF, line);
100104
}
@@ -168,32 +172,68 @@ private void buildSmallMoleculeReference(Model model, Map<String, String> chebiE
168172
String name = matcher.group(1);
169173
if (sy.contains("IUPAC_NAME")) {
170174
smr.setStandardName(name);
171-
} else if (sy.contains("InChIKey")) {
175+
} else if (sy.contains("InChIKey")) { //not the case with new chebi.obo - moved to _PROPERTY_VALUE, e.g.: property_value: http://purl.obolibrary.org/obo/chebi/inchikey "CRPUJAZIXJMDBK-DTWKUNHWSA-N" xsd:string
172176
if (name.startsWith("InChIKey=")) {
173177
//exclude the prefix
174178
name = name.substring(9);
175179
}
176180
//add RX because a InChIKey can map to several CHEBI IDs
177181
RelationshipXref rx = CPathUtils
178-
.findOrCreateRelationshipXref(RelTypeVocab.IDENTITY, "InChIKey", name, model);
182+
.findOrCreateRelationshipXref(RelTypeVocab.IDENTITY, "InChIKey", name, model);
179183
smr.addXref(rx);
180184
} else if (sy.contains("InChI=")) {
181185
String structureUri = Normalizer
182-
.uri(xmlBase, null, name, ChemicalStructure.class);
186+
.uri(xmlBase, null, name, ChemicalStructure.class);
183187
ChemicalStructure structure = (ChemicalStructure) model.getByID(structureUri);
184188
if (structure == null) {
185189
structure = model.addNew(ChemicalStructure.class, structureUri);
186190
structure.setStructureFormat(StructureFormatType.InChI);
187191
structure.setStructureData(name); //contains "InChI=" prefix
188192
}
189193
smr.setStructure(structure);
190-
} else if (sy.contains("FORMULA")) {
194+
} else if (sy.contains("FORMULA")) {//in new chebi.obo, this also moved from synonym:... to e.g. property_value: http://purl.obolibrary.org/obo/chebi/formula "C10H16" xsd:string
191195
smr.setChemicalFormula(name);
192196
smr.addName(name); //helps to map/search by name
193-
} else if (sy.contains("MASS")) {
197+
} else if (sy.contains("MASS")) {//in new chebi.obo, this also moved from synonym:... to e.g. property_value: http://purl.obolibrary.org/obo/chebi/mass "136.23404" xsd:string
194198
smr.setMolecularWeight(Float.parseFloat(name));
195199
} else {
196-
smr.addName(name); //incl. for SMILES
200+
smr.addName(name); //incl. for SMILES (in older chebi.obo data)
201+
}
202+
}
203+
}
204+
205+
//use property_value data to add names, structure, formula, InChIKey rel.xref, if the field is present
206+
final String propvals = chebiEntryMap.get(_PROPERTY_VALUE);
207+
if(propvals != null && !propvals.isEmpty()) {
208+
String[] entries = propvals.split("\t");
209+
for (String sy : entries) {
210+
Matcher matcher = namePattern.matcher(sy);
211+
if (!matcher.find()) {
212+
throw new IllegalStateException("Pattern failed to find a quoted text within: " + sy);
213+
}
214+
String name = matcher.group(1);
215+
if (sy.contains("purl.obolibrary.org/obo/chebi/inchikey")) {//e.g.: property_value: http://purl.obolibrary.org/obo/chebi/inchikey "CRPUJAZIXJMDBK-DTWKUNHWSA-N" xsd:string
216+
if (name.startsWith("InChIKey=")) {
217+
name = name.substring(9);
218+
}
219+
RelationshipXref rx = CPathUtils.findOrCreateRelationshipXref(RelTypeVocab.IDENTITY, "InChIKey", name, model);
220+
smr.addXref(rx);
221+
} else if (sy.contains("InChI=")) { //e.g. http://purl.obolibrary.org/obo/chebi/inchi "InChI=1S/C10H16/c1-7-8-4-5-9(6-8)10(7,2)3/h8-9H,1,4-6H2,2-3H3/t8-,9+/m0/s1" xsd:string
222+
String structureUri = Normalizer.uri(xmlBase, null, name, ChemicalStructure.class);
223+
ChemicalStructure structure = (ChemicalStructure) model.getByID(structureUri);
224+
if (structure == null) {
225+
structure = model.addNew(ChemicalStructure.class, structureUri);
226+
structure.setStructureFormat(StructureFormatType.InChI);
227+
structure.setStructureData(name); //keep "InChI=" prefix
228+
}
229+
smr.setStructure(structure);
230+
} else if (sy.contains("purl.obolibrary.org/obo/chebi/formula")) {//e.g. property_value: http://purl.obolibrary.org/obo/chebi/formula "C10H16" xsd:string
231+
smr.setChemicalFormula(name);
232+
smr.addName(name); //helps to map/search by name
233+
} else if (sy.contains("purl.obolibrary.org/obo/chebi/mass")) {//e.g. property_value: http://purl.obolibrary.org/obo/chebi/mass "136.23404" xsd:string
234+
smr.setMolecularWeight(Float.parseFloat(name));
235+
} else if (sy.contains("purl.obolibrary.org/obo/chebi/smiles")) {//e.g. property_value: http://purl.obolibrary.org/obo/chebi/smiles "CC1(C)[C@@H]2CC[C@@H](C2)C1=C" xsd:string
236+
smr.addName(name);
197237
}
198238
}
199239
}

‎src/main/java/cpath/service/BiopaxConverter.java

Lines changed: 1 addition & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -18,6 +18,7 @@
1818
import org.biopax.paxtools.io.*;
1919
import org.biopax.paxtools.model.*;
2020
import org.biopax.paxtools.model.level3.Provenance;
21+
import org.biopax.paxtools.normalizer.ConfigurableIDFetcher;
2122
import org.biopax.paxtools.pattern.miner.*;
2223
import org.biopax.paxtools.pattern.util.Blacklist;
2324
import org.slf4j.Logger;

‎src/main/java/cpath/web/BasicController.java

Lines changed: 10 additions & 39 deletions
Original file line numberDiff line numberDiff line change
@@ -9,8 +9,8 @@
99
import java.util.Arrays;
1010
import java.util.Set;
1111

12-
import jakarta.json.Json;
13-
import jakarta.json.JsonObjectBuilder;
12+
import com.fasterxml.jackson.databind.ObjectMapper;
13+
import com.fasterxml.jackson.databind.node.ObjectNode;
1414
import jakarta.servlet.http.HttpServletRequest;
1515
import jakarta.servlet.http.HttpServletResponse;
1616

@@ -45,6 +45,8 @@ public abstract class BasicController
4545

4646
protected Service service;
4747

48+
protected ObjectMapper jsonObjectMapper = new ObjectMapper();
49+
4850
@Autowired
4951
public void setService(Service service) {
5052
this.service = service;
@@ -186,55 +188,24 @@ final BufferedImage scaleImage(BufferedImage img, int width, int height)
186188
return newImage;
187189
}
188190

189-
190-
/*
191-
* Extracts the client's IP from the request headers.
192-
*/
193-
// private static String clientIpAddress(HttpServletRequest request)
194-
// {
195-
// String ip = request.getHeader("X-Forwarded-For");
196-
// if (ip == null || ip.isEmpty() || "unknown".equalsIgnoreCase(ip)) {
197-
// ip = request.getHeader("Proxy-Client-IP");
198-
// }
199-
// if (ip == null || ip.isEmpty() || "unknown".equalsIgnoreCase(ip)) {
200-
// ip = request.getHeader("WL-Proxy-Client-IP");
201-
// }
202-
// if (ip == null || ip.isEmpty() || "unknown".equalsIgnoreCase(ip)) {
203-
// ip = request.getHeader("HTTP_CLIENT_IP");
204-
// }
205-
// if (ip == null || ip.isEmpty() || "unknown".equalsIgnoreCase(ip)) {
206-
// ip = request.getHeader("HTTP_X_FORWARDED_FOR");
207-
// }
208-
// if (ip == null || ip.isEmpty() || "unknown".equalsIgnoreCase(ip)) {
209-
// ip = request.getRemoteAddr();
210-
// }
211-
// return ip;
212-
// }
213-
214191
void audit(HttpServletRequest request, ServiceQuery command, Set<String> providers, ErrorResponse err)
215192
{
216-
JsonObjectBuilder jb = Json.createObjectBuilder();
217-
218-
//get user-agent, IP, status, etc. from nginx/apache logs instead of here in the app...
219-
// jb.add("ip", clientIpAddress(request));
193+
ObjectNode root = jsonObjectMapper.createObjectNode();
220194

221195
if (err != null) {
222-
jb.add("error", err.toString());
196+
root.put("error", err.toString());
223197
}
224198

225199
if(command != null) {
226-
// TODO: change if there is any use (now we just add truncated string, not json object here
227-
//(can be very large if many URIs or SIF patterns are submitted in the request)
228-
jb.add("query", StringUtils.truncate(command.toString(),128));
200+
//truncate as it can be very large when many URIs or SIF patterns were submitted with the request
201+
root.put("query", StringUtils.truncate(command.toString(),128));
229202
}
230203

231204
if (!CollectionUtils.isEmpty(providers)) {
232-
jb.add("pro", Json.createArrayBuilder(providers));
205+
root.put("pro", jsonObjectMapper.valueToTree(providers));
233206
}
234207

235-
// jb.add("accept", request.getHeader("Accept"));
236-
237-
log.info(jb.build().toString());
208+
log.info(root.toString());
238209
}
239210

240211
}

‎src/main/resources/metadata.json

Lines changed: 3 additions & 3 deletions
Original file line numberDiff line numberDiff line change
@@ -4,7 +4,7 @@
44
"datasources": [
55
{
66
"dataUrl": "classpath:test_uniprot_data.dat.zip",
7-
"identifier": "TEST_UNIPROT",
7+
"identifier": "TESTUNIPROT",
88
"homepageUrl": "http://www.uniprot.org",
99
"name": [
1010
"UniProt"
@@ -18,13 +18,13 @@
1818
},
1919
{
2020
"dataUrl": "classpath:chebi.obo.zip",
21-
"identifier": "TEST_CHEBI",
21+
"identifier": "TESTCHEBI",
2222
"homepageUrl": "https://www.ebi.ac.uk/chebi/",
2323
"name": [
2424
"ChEBI"
2525
],
2626
"converterClass": "cpath.converter.ChebiOboConverter",
27-
"description": "Test ChEBI OBO",
27+
"description": "Test ChEBI OBO (extracted from chebi.obo release v231 and modified for integration/merge tests!)",
2828
"iconUrl": "https://pathwaycommons.github.io/cpath2/logos/chebi.png",
2929
"availability": "free",
3030
"type": "WAREHOUSE",

‎src/test/java/cpath/converter/ChebiConvertersTest.java

Lines changed: 10 additions & 8 deletions
Original file line numberDiff line numberDiff line change
@@ -62,7 +62,7 @@ public void convertObo() throws IOException {
6262

6363
// check some props
6464
assertEquals("(S)-lactic acid", smallMoleculeReference.getDisplayName());
65-
assertEquals(13, smallMoleculeReference.getName().size()); //now includes Wikipedia, SMILES(CHEBI) names
65+
assertEquals(13, smallMoleculeReference.getName().size()); //includes Wikipedia, SMILES names
6666
assertEquals("C3H6O3", smallMoleculeReference.getChemicalFormula());
6767
int relationshipXrefCount = 0;
6868
int unificationXrefCount = 0;
@@ -78,27 +78,29 @@ public void convertObo() throws IOException {
7878

7979
// following checks work in this test only (using in-memory model); with DAO - use getObject...
8080
assertTrue(model.containsID("http://bioregistry.io/chebi:20"));
81-
EntityReference er20 = (EntityReference) model.getByID("http://bioregistry.io/chebi:20");
81+
SmallMoleculeReference er20 = (SmallMoleculeReference) model.getByID("http://bioregistry.io/chebi:20");
8282
assertTrue(model.containsID("http://bioregistry.io/chebi:28"));
83-
// EntityReference er28 = (EntityReference) model.getByID("http://bioregistry.io/chebi:28");
8483
assertTrue(model.containsID("http://bioregistry.io/chebi:422"));
8584
EntityReference er422 = (EntityReference) model.getByID("http://bioregistry.io/chebi:422");
8685

8786
assertTrue(er20.getMemberEntityReferenceOf().isEmpty());
87+
assertNotNull(er20.getStructure());
88+
assertTrue(StringUtils.isNotBlank(er20.getChemicalFormula()));
89+
8890
assertTrue(er422.getMemberEntityReferenceOf().isEmpty());
89-
assertTrue(model.containsID("RX_chebi_CHEBI_422_multiple_parent_reference"));
9091

91-
// check new elements (created by the OBO converter) exist in the model;
92-
// (particularly, these assertions are important to test within the persistent model (DAO) session)
92+
assertTrue(model.containsID("RX_chebi_CHEBI_422_multiple_parent_reference"));
9393
assertTrue(model.containsID("RX_chebi_CHEBI_20_see-also"));
9494
assertTrue(model.containsID("RX_chebi_CHEBI_422_see-also"));
9595

96-
9796
//after refactoring, make sure there are no CHEBI:* xref.id anymore (only unprefixed)
9897
model.getObjects(Xref.class).stream().filter(x -> StringUtils.contains(x.getUri(),"chebi"))
9998
.forEach(x -> assertTrue(StringUtils.containsIgnoreCase(x.getId(),"CHEBI:")));
100-
10199
model.getObjects(Xref.class).stream().filter(x -> StringUtils.contains(x.getUri(),"chebi"))
102100
.forEach(x -> assertEquals("chebi", x.getDb()));
101+
102+
SmallMoleculeReference smr28 = (SmallMoleculeReference) model.getByID("http://bioregistry.io/chebi:28");
103+
assertNotNull(smr28);
104+
assertEquals(8, smr28.getXref().size());
103105
}
104106
}

‎src/test/java/cpath/service/BiopaxConverterTest.java

Lines changed: 29 additions & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -5,16 +5,23 @@
55
import cpath.service.api.OutputFormat;
66
import cpath.service.jaxb.DataResponse;
77
import cpath.service.jaxb.ServiceResponse;
8+
import org.biopax.paxtools.io.SimpleIOHandler;
89
import org.biopax.paxtools.model.BioPAXLevel;
910
import org.biopax.paxtools.model.Model;
1011
import org.biopax.paxtools.model.level3.Pathway;
12+
import org.junit.jupiter.api.Assertions;
1113
import org.junit.jupiter.api.Test;
1214

15+
import java.io.IOException;
16+
import java.net.URI;
17+
import java.nio.file.Files;
18+
import java.nio.file.Paths;
19+
1320

1421
public class BiopaxConverterTest {
1522

1623
@Test
17-
public final void testToJsonLd() {
24+
public final void testToJsonld() {
1825
Model m = BioPAXLevel.L3.getDefaultFactory().createModel();
1926
m.setXmlBase("http://pathwaycommons.org/pc2/"); //Jena fails if not set!
2027
Pathway bpe = m.addNew(Pathway.class, "http://pathwaycommons.org/pc2/Pathway_test-URI");
@@ -25,4 +32,25 @@ public final void testToJsonLd() {
2532
assertFalse(sr.isEmpty());
2633
}
2734

35+
@Test
36+
public final void testDemo1ToJsonld() throws IOException {
37+
Model m = new SimpleIOHandler().convertFromOWL(getClass().getResourceAsStream("/demo-pathway.owl"));
38+
ServiceResponse sr = new BiopaxConverter(null).convert(m, OutputFormat.JSONLD, null);
39+
40+
assertTrue(sr instanceof DataResponse && !sr.isEmpty());
41+
42+
String resf = ((DataResponse)sr).getData().toString(); //must be a temp. file name
43+
String res = Files.readString(Paths.get(resf));
44+
45+
Assertions.assertAll(
46+
() -> Assertions.assertThrows(IllegalArgumentException.class, () -> URI.create("http://")), //bad URI
47+
() -> Assertions.assertTrue(res.contains("@id\" : \"http://bioregistry.io/chebi:20")),
48+
() -> Assertions.assertTrue(res.contains("@id\" : \"http://bioregistry.io/mi:0361")),//as long as it has 'http://' (valid abs. uri w/o schema would fail here due Jena bug)
49+
() -> Assertions.assertTrue(res.contains("@id\" : \"chebi:20")), //CURIE of a standard/normalized SMR's UnificationXref
50+
() -> Assertions.assertTrue(res.contains("\"@id\" : \"http://www.biopax.org/release/biopax-level3.owl#displayName\"")),
51+
() -> Assertions.assertTrue(res.contains("\"displayName\" : \"(+)-camphene\"")), //chebi:20
52+
() -> Assertions.assertTrue(res.contains("\"id\" : \"CHEBI:20\"")) //with jena v3 (e.g. 3.2.0 or 3.17.0)
53+
);
54+
}
55+
2856
}

‎src/test/java/cpath/service/CPathUtilsTest.java

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -74,7 +74,7 @@ public void readWriteMetadata() {
7474
Metadata metadata = CPathUtils.readMetadata(url);
7575
List<Datasource> datasources = metadata.getDatasources();
7676
final Datasource datasource = datasources.stream()
77-
.filter(m -> m.getIdentifier().equals("TEST_UNIPROT"))
77+
.filter(m -> m.getIdentifier().equals("TESTUNIPROT"))
7878
.findFirst().orElse(null);
7979
assertAll(
8080
() -> assertEquals(3, datasources.size()),

‎src/test/java/cpath/service/ConsoleApplicationIT.java

Lines changed: 2 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -182,9 +182,9 @@ public void premergeAndMerge() throws IOException {
182182
assertTrue(service.settings().getOrganismsAsTaxonomyToNameMap().containsKey("9606"));
183183
assertEquals("Homo sapiens", service.settings().getOrganismsAsTaxonomyToNameMap().get("9606"));
184184

185-
Datasource ds = service.metadata().findByIdentifier("TEST_UNIPROT");
185+
Datasource ds = service.metadata().findByIdentifier("TESTUNIPROT");
186186
assertNotNull(ds);
187-
ds = service.metadata().findByIdentifier("TEST_CHEBI");
187+
ds = service.metadata().findByIdentifier("TESTCHEBI");
188188
assertNotNull(ds);
189189
ds = service.metadata().findByIdentifier("TEST_MAPPING");
190190
assertNotNull(ds);

‎src/test/resources/chebi.obo.zip

696 Bytes
Binary file not shown.

0 commit comments

Comments
 (0)