added parallelizations

added more efficient indexation
added beginning of extended annotation of molecules

src/main/java/de/unijena/cheminf/npopensourcecollector/NpOpenSourceCollectorApplication.java

@@ -50,11 +50,11 @@ public void run(String... args) throws Exception {
 
 
         //cleaning the DB before filling it
-        //mongoTemplate.getDb().drop();
+        mongoTemplate.getDb().drop();
 
 
 
-        System.out.println("Code version from 20 August 2019");
+        System.out.println("Code version from 23rd september 2019");
 
         if (args.length > 0) {
             String dataDirectory = args[0];
@@ -87,10 +87,10 @@ public void run(String... args) throws Exception {
 
 
 
-                updaterService.updateSourceNaturalProducts();
                 //compute similarities between natural products
                 similarityComputationService.generateAllPairs();
-                similarityComputationService.computeSimilarities();
+                //similarityComputationService.computeSimilarities();
+                similarityComputationService.doParallelizedWork();
 
 
 

src/main/java/de/unijena/cheminf/npopensourcecollector/misc/DatabaseTypeChecker.java

@@ -0,0 +1,95 @@
+package de.unijena.cheminf.npopensourcecollector.misc;
+
+import org.springframework.stereotype.Service;
+
+import java.util.Arrays;
+import java.util.HashSet;
+
+
+@Service
+public class DatabaseTypeChecker {
+
+
+
+
+        private final String[] africa = {"afrodb", "afrocancer", "afromalariadb", "afrotryp", "conmednp", "etm", "mitishamba", "nanpdb", "p-anapl", "sancdb"};
+        private final String[] china = {"him", "hit", "tcmdb_taiwan", "tcmid", "tipdb"};
+        private final String[] india = {"imppat", "inpacdb"};
+        private final String[] europe = {"tppt"};
+        private final String[] america = {"nubbedb", "uefs", "biofacquim"};
+
+        private final HashSet<String> continentAfrica = new HashSet<String>(Arrays.asList(africa));
+        private final HashSet<String> continentIndia = new HashSet<String>(Arrays.asList(india));
+        private final HashSet<String> continentChina = new HashSet<String>(Arrays.asList(china));
+        private final HashSet<String> continentEurope = new HashSet<String>(Arrays.asList(europe));
+        private final HashSet<String> continentAmerica = new HashSet<String>(Arrays.asList(america));
+
+
+
+        private final String[] plants = {"uefs","tppt","tmdb","tipdb","tcmid", "tcmdb_taiwan","spektraris","sancdb",
+                "respect","p-anapl", "npact","nanpdb","mitishamba","inpacdb","imppat", "hit","him","etm","conmednp",
+                "afrotryp", "afromalariadb","afrocancer","afrodb"};
+        private final String[] bacteria = {"streptomedb"};
+        private final String[] fungi = {"lichendatabase"};
+        private final String[] animals = {};
+        private final String[] marine = {};
+        private final String[] mixed = {"nubbedb","npcare","npatlas","npass","analyticon_all_np", "biofacquim"};
+
+        private final HashSet<String> taxPlants = new HashSet<String>(Arrays.asList(plants));
+        private final HashSet<String> taxBacteria = new HashSet<String>(Arrays.asList(bacteria));
+        private final HashSet<String> taxFungi = new HashSet<String>(Arrays.asList(fungi));
+        private final HashSet<String> taxAnimals = new HashSet<String>(Arrays.asList(animals));
+        private final HashSet<String> taxMarine = new HashSet<String>(Arrays.asList(marine));
+        private final HashSet<String> taxMixed = new HashSet<String>(Arrays.asList(mixed));
+
+
+        public  String checkContinent(String sourceDB){
+
+            if(continentAfrica.contains(sourceDB)){
+                return "africa";
+            }
+            else if(continentChina.contains(sourceDB)){
+                return "china";
+            }
+            else if(continentIndia.contains(sourceDB)){
+                return "india";
+            }
+            else if(continentEurope.contains(sourceDB)){
+                return "europe";
+            }
+            else if(continentAmerica.contains(sourceDB)){
+                return "southamerica";
+            }
+            else {
+                return "nogeo";
+            }
+        }
+
+
+        public  String checkKingdom(String sourceDB){
+            if(taxPlants.contains(sourceDB)){
+                return "plants";
+            }
+            else if(taxBacteria.contains(sourceDB)){
+                return "bacteria";
+            }
+            else if(taxAnimals.contains(sourceDB)){
+                return "animals";
+            }
+            else if(taxFungi.contains(sourceDB)){
+                return "fungi";
+            }
+            else if(taxMarine.contains(sourceDB)){
+                return "marine";
+            }
+            else if(taxMixed.contains(sourceDB)){
+                return "mixed";
+            }
+            else{
+                return "notax";
+            }
+
+        }
+
+
+}

src/main/java/de/unijena/cheminf/npopensourcecollector/misc/MoleculeChecker.java

@@ -108,7 +108,6 @@ public IAtomContainer checkMolecule(IAtomContainer molecule){
             }
 
 
-            //Remove aromaticity
             String smi;
             SmilesGenerator sg = new SmilesGenerator(SmiFlavor.Unique);
             SmilesParser sp = new SmilesParser(DefaultChemObjectBuilder.getInstance());

src/main/java/de/unijena/cheminf/npopensourcecollector/mongocollections/NPDatabase.java

@@ -0,0 +1,66 @@
+package de.unijena.cheminf.npopensourcecollector.mongocollections;
+
+import org.springframework.data.annotation.Id;
+import org.springframework.data.mongodb.core.mapping.Document;
+
+@Document
+public class NPDatabase {
+
+    @Id
+    public String id;
+
+    String name;
+
+    String localFileName;
+
+    String url;
+
+    String comments;
+
+    Integer nb_unique_molecules;
+
+
+    public String getId() {
+        return id;
+    }
+
+    public String getName() {
+        return name;
+    }
+
+    public void setName(String name) {
+        this.name = name;
+    }
+
+    public String getUrl() {
+        return url;
+    }
+
+    public void setUrl(String url) {
+        this.url = url;
+    }
+
+    public String getComments() {
+        return comments;
+    }
+
+    public void setComments(String comments) {
+        this.comments = comments;
+    }
+
+    public Integer getNb_unique_molecules() {
+        return nb_unique_molecules;
+    }
+
+    public void setNb_unique_molecules(Integer nb_unique_molecules) {
+        this.nb_unique_molecules = nb_unique_molecules;
+    }
+
+    public String getLocalFileName() {
+        return localFileName;
+    }
+
+    public void setLocalFileName(String localFileName) {
+        this.localFileName = localFileName;
+    }
+}

src/main/java/de/unijena/cheminf/npopensourcecollector/mongocollections/NPDatabaseRepository.java

@@ -0,0 +1,6 @@
+package de.unijena.cheminf.npopensourcecollector.mongocollections;
+
+import org.springframework.data.mongodb.repository.MongoRepository;
+
+public interface NPDatabaseRepository extends MongoRepository<NPDatabase, String> {
+}

src/main/java/de/unijena/cheminf/npopensourcecollector/mongocollections/NPSimilarity.java

@@ -9,9 +9,9 @@ public class NPSimilarity {
     @Id
     public String id;
 
-    public UniqueNaturalProduct uniqueNaturalProduct1;
+    public String uniqueNaturalProductID1;
 
-    public UniqueNaturalProduct uniqueNaturalProduct2;
+    public String uniqueNaturalProductID2;
 
     public Double tanimoto;
 
@@ -26,24 +26,24 @@ public void setId(String id) {
         this.id = id;
     }
 
-    public UniqueNaturalProduct getUniqueNaturalProduct1() {
-        return uniqueNaturalProduct1;
+    public Double getTanimoto() {
+        return tanimoto;
     }
 
-    public void setUniqueNaturalProduct1(UniqueNaturalProduct uniqueNaturalProduct1) {
-        this.uniqueNaturalProduct1 = uniqueNaturalProduct1;
+    public String getUniqueNaturalProductID1() {
+        return uniqueNaturalProductID1;
     }
 
-    public UniqueNaturalProduct getUniqueNaturalProduct2() {
-        return uniqueNaturalProduct2;
+    public void setUniqueNaturalProductID1(String uniqueNaturalProductID1) {
+        this.uniqueNaturalProductID1 = uniqueNaturalProductID1;
     }
 
-    public void setUniqueNaturalProduct2(UniqueNaturalProduct uniqueNaturalProduct2) {
-        this.uniqueNaturalProduct2 = uniqueNaturalProduct2;
+    public String getUniqueNaturalProductID2() {
+        return uniqueNaturalProductID2;
     }
 
-    public Double getTanimoto() {
-        return tanimoto;
+    public void setUniqueNaturalProductID2(String uniqueNaturalProductID2) {
+        this.uniqueNaturalProductID2 = uniqueNaturalProductID2;
     }
 
     public void setTanimoto(Double tanimoto) {

src/main/java/de/unijena/cheminf/npopensourcecollector/mongocollections/SourceNaturalProduct.java

@@ -4,13 +4,16 @@
 import org.springframework.data.mongodb.core.index.Indexed;
 import org.springframework.data.mongodb.core.mapping.Document;
 
+import java.util.ArrayList;
+
 
 @Document
 public class SourceNaturalProduct {
 
     @Id
     public String id;
 
+    @Indexed
     public String source;
 
     public String originalSmiles;
@@ -21,8 +24,10 @@ public class SourceNaturalProduct {
 
     public String originalInchiKey;
 
+    @Indexed
     public String simpleInchi;
 
+    @Indexed
     public String simpleInchiKey;
 
     public Integer heavyAtomNumber;
@@ -37,6 +42,20 @@ public class SourceNaturalProduct {
 
     public UniqueNaturalProduct uniqueNaturalProduct;
 
+    public ArrayList<String> citation;
+
+    public ArrayList<Integer> taxid;
+
+    public ArrayList<String> organismText;
+
+    public String continent;
+
+    public ArrayList<String> geographicLocation;
+
+    public String name;
+
+    public ArrayList<String> synonyms;
+
 
 
 
@@ -162,4 +181,60 @@ public UniqueNaturalProduct getUniqueNaturalProduct() {
     public void setUniqueNaturalProduct(UniqueNaturalProduct uniqueNaturalProduct) {
         this.uniqueNaturalProduct = uniqueNaturalProduct;
     }
+
+    public ArrayList<String> getCitation() {
+        return citation;
+    }
+
+    public void setCitation(ArrayList<String> citation) {
+        this.citation = citation;
+    }
+
+    public ArrayList<Integer> getTaxid() {
+        return taxid;
+    }
+
+    public void setTaxid(ArrayList<Integer> taxid) {
+        this.taxid = taxid;
+    }
+
+    public String getContinent() {
+        return continent;
+    }
+
+    public void setContinent(String continent) {
+        this.continent = continent;
+    }
+
+    public ArrayList<String> getOrganismText() {
+        return organismText;
+    }
+
+    public void setOrganismText(ArrayList<String> organismText) {
+        this.organismText = organismText;
+    }
+
+    public String getName() {
+        return name;
+    }
+
+    public void setName(String name) {
+        this.name = name;
+    }
+
+    public ArrayList<String> getSynonyms() {
+        return synonyms;
+    }
+
+    public void setSynonyms(ArrayList<String> synonyms) {
+        this.synonyms = synonyms;
+    }
+
+    public ArrayList<String> getGeographicLocation() {
+        return geographicLocation;
+    }
+
+    public void setGeographicLocation(ArrayList<String> geographicLocation) {
+        this.geographicLocation = geographicLocation;
+    }
 }