Merge branch 'feature/SBML_scraping' into 'develop'

Expand Toolbox with SBML scrapping app

See merge request !9

met4j-toolbox/src/main/java/fr/inrae/toulouse/metexplore/met4j_toolbox/attributes/ExtractPathways.java

+package fr.inrae.toulouse.metexplore.met4j_toolbox.attributes;
+
+import fr.inrae.toulouse.metexplore.met4j_core.biodata.*;
+import fr.inrae.toulouse.metexplore.met4j_core.biodata.collection.BioCollection;
+import fr.inrae.toulouse.metexplore.met4j_io.jsbml.reader.JsbmlReader;
+import fr.inrae.toulouse.metexplore.met4j_io.jsbml.reader.Met4jSbmlReaderException;
+import fr.inrae.toulouse.metexplore.met4j_io.jsbml.writer.JsbmlWriter;
+import fr.inrae.toulouse.metexplore.met4j_io.jsbml.writer.Met4jSbmlWriterException;
+import fr.inrae.toulouse.metexplore.met4j_toolbox.generic.AbstractMet4jApplication;
+import org.kohsuke.args4j.Option;
+
+import java.io.IOException;
+
+public class ExtractPathways extends AbstractMet4jApplication {
+
+    @Option(name = "-i", usage = "input SBML file", required = true)
+    public String inputPath = null;
+
+    @Option(name = "-o", usage = "output SBML file", required = true)
+    public String outputPath = null;
+
+    @Option(name = "-p", usage = "pathway identifiers, separated by \"+\" sign if more than one", required = true)
+    public String pathwayId = null;
+
+    public static void main(String[] args) throws IOException, Met4jSbmlReaderException, Met4jSbmlWriterException {
+
+        ExtractPathways app = new ExtractPathways();
+
+        app.parseArguments(args);
+
+        app.run();
+
+    }
+
+
+    public void run() throws IOException, Met4jSbmlReaderException, Met4jSbmlWriterException {
+        //read smbl
+        JsbmlReader reader = new JsbmlReader(this.inputPath);
+        BioNetwork network = reader.read();
+        System.out.println("Number of reactions in original network: "+network.getReactionsView().size());
+        System.out.println("Number of species in original network: "+network.getMetabolitesView().size());
+        System.out.println("Number of genes in original network: "+network.getGenesView().size());
+
+
+        //get all reactions & metabolites
+        BioCollection<BioReaction> reactions = new BioCollection<>(network.getReactionsView());
+        BioCollection<BioMetabolite> metabolites = new BioCollection<>(network.getMetabolitesView());
+        BioCollection<BioGene> genes = new BioCollection<>(network.getGenesView());
+
+        //get pathways
+        BioCollection<BioPathway> pathways = new BioCollection<>();
+        for(String id : pathwayId.split("\\+")){
+            BioPathway pathway = network.getPathwaysView().get(id);
+            if(pathway!=null){
+                pathways.add(pathway);
+                System.out.println("Number of reactions in pathway "+pathway.getName()+" ("+id+"): "+network.getReactionsFromPathways(pathway).size());
+                System.out.println("Number of species in pathway "+pathway.getName()+" ("+id+"): "+network.getMetabolitesFromPathway(pathway).size());
+                System.out.println("Number of genes in pathway "+pathway.getName()+" ("+id+"): "+network.getGenesFromPathways(pathway).size());
+            }else{
+                System.out.println("Error: Pathway "+id+" not found in network, please check sbml file.");
+            }
+        }
+
+        //remove pathway's reactions and metabolites from list
+        BioCollection<BioReaction> pathwaysReactions = network.getReactionsFromPathways(pathways);
+        reactions.removeAll(pathwaysReactions);
+        metabolites.removeAll(network.getMetabolitesFromReactions(pathwaysReactions));
+        genes.removeAll(network.getGenesFromReactions(pathwaysReactions));
+
+        //remove remaining reactions
+        network.removeOnCascade(reactions);
+        network.removeOnCascade(metabolites);
+        network.removeOnCascade(genes);
+        System.out.println("Number of reactions in network: "+network.getReactionsView().size());
+        System.out.println("Number of species in network: "+network.getMetabolitesView().size());
+        System.out.println("Number of genes in network: "+network.getGenesView().size());
+
+        //export network
+        JsbmlWriter w = new JsbmlWriter(outputPath, network);
+        w.write();
+        System.err.println("network exported.");
+        return;
+    }
+
+    @Override
+    public String getLabel() {
+        return this.getClass().getSimpleName();
+    }
+
+    @Override
+    public String getLongDescription() {
+        return "\"Extract pathway(s) from GSMN: From a SBML file, Create a sub-network SBML file including only a selection of pathways";
+    }
+
+    @Override
+    public String getShortDescription() {
+        return "Extract pathway(s) from GSMN";
+    }
+}

met4j-toolbox/src/main/java/fr/inrae/toulouse/metexplore/met4j_toolbox/attributes/ExtractSbmlAnnot.java

+package fr.inrae.toulouse.metexplore.met4j_toolbox.attributes;
+
+import fr.inrae.toulouse.metexplore.met4j_core.biodata.BioEntity;
+import fr.inrae.toulouse.metexplore.met4j_core.biodata.BioNetwork;
+import fr.inrae.toulouse.metexplore.met4j_core.biodata.BioRef;
+import fr.inrae.toulouse.metexplore.met4j_core.biodata.collection.BioCollection;
+import fr.inrae.toulouse.metexplore.met4j_io.jsbml.reader.JsbmlReader;
+import fr.inrae.toulouse.metexplore.met4j_io.jsbml.reader.Met4jSbmlReaderException;
+import fr.inrae.toulouse.metexplore.met4j_toolbox.generic.AbstractMet4jApplication;
+import org.kohsuke.args4j.Option;
+
+import java.io.FileWriter;
+import java.io.IOException;
+import java.util.*;
+
+public class ExtractSbmlAnnot extends AbstractMet4jApplication {
+
+    @Option(name = "-i", usage = "input SBML file", required = true)
+    public String inputPath = null;
+
+    @Option(name = "-o", usage = "output file path", required = true)
+    public String outputPath = null;
+
+    enum entity { METABOLITE,REACTION,GENE}
+
+    @Option(name="-export", usage = "the type of entity to extract annotation, either metabolite, reaction, or gene", required = true)
+    public entity export;
+
+    @Option(name="-db", usage = "name of the referenced database to export annotations from, as listed in notes or identifiers.org base uri", required = true)
+    public String db;
+
+    @Option(name="-uniq", usage = "keep only one identifier if multiple are referenced for the same entity", required = false)
+    public Boolean uniq = false;
+
+    @Option(name="-skip", usage = "Skip entities without the selected annotations, by default output them with NA value", required = false)
+    public Boolean skip = false;
+
+    public String sep = "\t";
+
+
+    public static void main(String[] args) throws IOException, Met4jSbmlReaderException {
+
+        ExtractSbmlAnnot app = new ExtractSbmlAnnot();
+
+        app.parseArguments(args);
+
+        app.run();
+
+    }
+
+    private void run() throws IOException, Met4jSbmlReaderException {
+        //open file
+        FileWriter fw = new FileWriter(outputPath);
+
+        //read smbl
+        JsbmlReader reader = new JsbmlReader(this.inputPath);
+        BioNetwork network = reader.read();
+
+        BioCollection<? extends BioEntity> entities = new BioCollection<>();
+        if(export==entity.METABOLITE){
+            entities=network.getMetabolitesView();
+        }else if(export==entity.REACTION){
+            entities=network.getReactionsView();
+        }else if(export==entity.GENE){
+            entities=network.getGenesView();
+        }
+
+        //write header
+        fw.write(export.name()+sep+db.toUpperCase()+"\n");
+
+        //export annotations
+                //keep track of successful export
+        int i = 0;
+        for(BioEntity e : entities){
+
+            Set<BioRef> refSet = e.getRefs(db);
+            if(refSet!=null){
+                i+=1;
+                if(uniq)refSet=  new HashSet<BioRef>(Arrays.asList(refSet.iterator().next()));
+                for(BioRef ref : refSet){
+                    StringBuffer sb = new StringBuffer();
+                    sb.append(e.getId());
+                    sb.append(sep);
+                    sb.append(ref.getId());
+                    sb.append("\n");
+                    fw.write(sb.toString());
+                }
+            }else if(!skip){
+                StringBuffer sb = new StringBuffer();
+                sb.append(e.getId());
+                sb.append(sep);
+                sb.append("NA\n");
+                fw.write(sb.toString());
+            }
+
+        }
+        fw.close();
+        System.out.println("annotations found for "+i+"/"+entities.size()+" "+export.name().toLowerCase()+((i>1)?"s":""));
+
+    }
+
+    @Override
+    public String getLabel() {
+        return this.getClass().getSimpleName();
+    }
+
+    @Override
+    public String getLongDescription() {
+        return "Extract databases' references from SBML annotations or notes. " +
+                "The references are exported as a tabulated file with one column with the SBML compound, " +
+                "reaction or gene identifiers, and one column with the corresponding database identifier." +
+                "The name of the targeted database need to be provided under the same form than the one used " +
+                "in the notes field or the identifiers.org uri";
+    }
+
+    @Override
+    public String getShortDescription() {
+        return "Extract databases' references from SBML annotations or notes.";
+    }
+}