From 09c23a938d68bd3993f88cc0a5ff31bb917e4948 Mon Sep 17 00:00:00 2001
From: cfrainay <clement.frainay@inrae.fr>
Date: Thu, 28 Jul 2022 16:44:43 +0200
Subject: [PATCH 1/3] Add app to turn SBML into lists of entities
---
.../attributes/DecomposeSBML.java | 134 ++++++++++++++++++
1 file changed, 134 insertions(+)
create mode 100644 met4j-toolbox/src/main/java/fr/inrae/toulouse/metexplore/met4j_toolbox/attributes/DecomposeSBML.java
diff --git a/met4j-toolbox/src/main/java/fr/inrae/toulouse/metexplore/met4j_toolbox/attributes/DecomposeSBML.java b/met4j-toolbox/src/main/java/fr/inrae/toulouse/metexplore/met4j_toolbox/attributes/DecomposeSBML.java
new file mode 100644
index 000000000..c483a3f9d
--- /dev/null
+++ b/met4j-toolbox/src/main/java/fr/inrae/toulouse/metexplore/met4j_toolbox/attributes/DecomposeSBML.java
@@ -0,0 +1,134 @@
+package fr.inrae.toulouse.metexplore.met4j_toolbox.attributes;
+
+import fr.inrae.toulouse.metexplore.met4j_core.biodata.*;
+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 fr.inrae.toulouse.metexplore.met4j_toolbox.generic.annotations.EnumFormats;
+import fr.inrae.toulouse.metexplore.met4j_toolbox.generic.annotations.EnumParameterTypes;
+import fr.inrae.toulouse.metexplore.met4j_toolbox.generic.annotations.Format;
+import fr.inrae.toulouse.metexplore.met4j_toolbox.generic.annotations.ParameterType;
+import org.kohsuke.args4j.Option;
+
+import java.io.FileWriter;
+import java.io.IOException;
+import java.io.PrintWriter;
+
+public class DecomposeSBML extends AbstractMet4jApplication {
+
+
+ @ParameterType(name= EnumParameterTypes.InputFile)
+ @Format(name= EnumFormats.Sbml)
+ @Option(name = "-i", usage = "Input SBML file", required = true)
+ public String sbml;
+
+ @ParameterType(name= EnumParameterTypes.Boolean)
+ @Option(name = "-m", aliases = {"--metabolites"}, usage = "Extract Metabolites", required = false)
+ public Boolean printMetabolites = false;
+ @ParameterType(name= EnumParameterTypes.Boolean)
+ @Option(name = "-r", aliases = {"--reactions"}, usage = "Extract Reactions", required = false)
+ public Boolean printReactions = false;
+ @ParameterType(name= EnumParameterTypes.Boolean)
+ @Option(name = "-c", aliases = {"--compartments"}, usage = "Extract Compartments", required = false)
+ public Boolean printCompartments = false;
+ @ParameterType(name= EnumParameterTypes.Boolean)
+ @Option(name = "-p", aliases = {"--pathways"}, usage = "Extract Pathways", required = false)
+ public Boolean printPathways = false;
+ @ParameterType(name= EnumParameterTypes.Boolean)
+ @Option(name = "-g", aliases = {"--genes"}, usage = "Extract Genes", required = false)
+ public Boolean printGenes = false;
+
+ @ParameterType(name= EnumParameterTypes.Boolean)
+ @Option(name = "-nt", aliases = {"--noTypeCol"}, usage = "Do not output type column", required = false)
+ public Boolean noTypeCol = false;
+
+ @ParameterType(name= EnumParameterTypes.OutputFile)
+ @Format(name= EnumFormats.Tsv)
+ @Option(name = "-o", usage = "Output file", required=true)
+ public String outputFile;
+
+ /**
+ * <p>main.</p>
+ *
+ * @param args an array of {@link java.lang.String} objects.
+ */
+ public static void main(String[] args) {
+ DecomposeSBML app = new DecomposeSBML();
+ app.parseArguments(args);
+ app.run();
+ }
+
+ /**
+ * <p>run.</p>
+ */
+ public void run() {
+
+ String fileIn = this.sbml;
+ JsbmlReader reader = new JsbmlReader(fileIn);
+ BioNetwork network = null;
+ try {
+ network = reader.read();
+ } catch (Met4jSbmlReaderException e) {
+ System.err.println("Error while reading the SBML file");
+ System.err.println(e.getMessage());
+ System.exit(1);
+ }
+
+ //default case: everything printed
+ if(!(printMetabolites|printReactions|printPathways|printGenes|printCompartments)){
+ printMetabolites=printReactions=printPathways=printGenes=printCompartments=true;
+ }
+
+ try (PrintWriter writer = new PrintWriter(new FileWriter(this.outputFile, false))) {
+
+ if(!noTypeCol) writer.println("ID\tEntity.Type");
+ if(printMetabolites){
+ for (BioMetabolite metabolite : network.getMetabolitesView()) {
+ writer.println(noTypeCol ? metabolite.getId() : metabolite.getId() + "\tMETABOLITE");
+ }
+ }
+ if(printReactions){
+ for (BioReaction reaction : network.getReactionsView()) {
+ writer.println(noTypeCol ? reaction.getId() : reaction.getId() + "\tREACTION");
+ }
+ }
+ if(printGenes){
+ for (BioGene gene : network.getGenesView()) {
+ writer.println(noTypeCol ? gene.getId() : gene.getId() + "\tGENE");
+ }
+ }
+ if(printPathways){
+ for (BioPathway pathway : network.getPathwaysView()) {
+ writer.println(noTypeCol ? pathway.getId() : pathway.getId() + "\tPATHWAY");
+ }
+ }
+ if(printCompartments){
+ for (BioCompartment compartment : network.getCompartmentsView()) {
+ writer.println(noTypeCol ? compartment.getId() : compartment.getId() + "\tCOMPARTMENT");
+ }
+ }
+
+ } catch (IOException e) {
+ System.err.println("Error while writing SBML entities");
+ System.exit(1);
+ }
+ }
+
+ @Override
+ public String getLabel() {
+ return this.getClass().getSimpleName();
+ }
+
+ @Override
+ public String getLongDescription() {
+ return "Parse SBML to render list of composing entities: metabolites, reactions, genes, pathways and compartments. " +
+ "The output file is a tsv with two columns, one with entities identifiers, and one with the entity type. " +
+ "If no entity type is selected, by default all of them are taken into account. " +
+ "Only identifiers are written, attributes can be extracted from dedicated apps or from the SBML2Tab.";
+ }
+
+ @Override
+ public String getShortDescription() {
+ return "Parse SBML to render list of composing entities: metabolites, reactions, genes and others.";
+ }
+}
--
GitLab
From 06239a8fc1f9a00956394da2e1291a094502d3e7 Mon Sep 17 00:00:00 2001
From: cfrainay <clement.frainay@inrae.fr>
Date: Thu, 28 Jul 2022 17:54:57 +0200
Subject: [PATCH 2/3] Add app to get substrates and/or products list from a
list of reactions
---
.../attributes/GetReactantsFromReactions.java | 141 ++++++++++++++++++
1 file changed, 141 insertions(+)
create mode 100644 met4j-toolbox/src/main/java/fr/inrae/toulouse/metexplore/met4j_toolbox/attributes/GetReactantsFromReactions.java
diff --git a/met4j-toolbox/src/main/java/fr/inrae/toulouse/metexplore/met4j_toolbox/attributes/GetReactantsFromReactions.java b/met4j-toolbox/src/main/java/fr/inrae/toulouse/metexplore/met4j_toolbox/attributes/GetReactantsFromReactions.java
new file mode 100644
index 000000000..336ac4dd7
--- /dev/null
+++ b/met4j-toolbox/src/main/java/fr/inrae/toulouse/metexplore/met4j_toolbox/attributes/GetReactantsFromReactions.java
@@ -0,0 +1,141 @@
+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_mapping.Mapper;
+import fr.inrae.toulouse.metexplore.met4j_toolbox.generic.AbstractMet4jApplication;
+import fr.inrae.toulouse.metexplore.met4j_toolbox.generic.annotations.EnumFormats;
+import fr.inrae.toulouse.metexplore.met4j_toolbox.generic.annotations.EnumParameterTypes;
+import fr.inrae.toulouse.metexplore.met4j_toolbox.generic.annotations.Format;
+import fr.inrae.toulouse.metexplore.met4j_toolbox.generic.annotations.ParameterType;
+import org.kohsuke.args4j.Option;
+
+import java.io.FileWriter;
+import java.io.IOException;
+import java.io.PrintWriter;
+
+public class GetReactantsFromReactions extends AbstractMet4jApplication {
+
+
+ @ParameterType(name= EnumParameterTypes.InputFile)
+ @Format(name= EnumFormats.Sbml)
+ @Option(name = "-i", usage = "Input SBML file", required = true)
+ public String sbml;
+
+ @ParameterType(name= EnumParameterTypes.InputFile)
+ @Format(name= EnumFormats.Tsv)
+ @Option(name = "-r", usage = "Input Reaction file", required = true)
+ public String reactionFile;
+
+ @ParameterType(name= EnumParameterTypes.Text)
+ @Option(name = "-sep", usage = "Separator in reaction file", required = false)
+ public String sep = "\t";
+ @ParameterType(name= EnumParameterTypes.Boolean)
+ @Option(name = "-header", usage = "Skip reaction file header", required = false)
+ public boolean hasHeader = false;
+ @ParameterType(name= EnumParameterTypes.Integer)
+ @Option(name = "-col", usage = "Column number in reaction file (first as 1)", required = false)
+ public int i=1;
+
+ @ParameterType(name= EnumParameterTypes.Boolean)
+ @Option(name = "-s", aliases = {"--substrates"}, usage = "Extract substrates only", required = false)
+ public Boolean printSubstrates = false;
+ @ParameterType(name= EnumParameterTypes.Boolean)
+ @Option(name = "-p", aliases = {"--products"}, usage = "Extract products only", required = false)
+ public Boolean printProducts = false;
+
+ @ParameterType(name= EnumParameterTypes.OutputFile)
+ @Format(name= EnumFormats.Tsv)
+ @Option(name = "-o", usage = "Output file", required=true)
+ public String outputFile;
+
+ /**
+ * <p>main.</p>
+ *
+ * @param args an array of {@link java.lang.String} objects.
+ */
+ public static void main(String[] args) {
+ GetReactantsFromReactions app = new GetReactantsFromReactions();
+ app.parseArguments(args);
+ app.run();
+ }
+
+
+ /**
+ * <p>run.</p>
+ */
+ public void run() {
+
+ //read SBML, create bionetwork
+ String fileIn = this.sbml;
+ JsbmlReader reader = new JsbmlReader(fileIn);
+ BioNetwork network = null;
+ try {
+ network = reader.read();
+ } catch (Met4jSbmlReaderException e) {
+ System.err.println("Error while reading the SBML file");
+ System.err.println(e.getMessage());
+ System.exit(1);
+ }
+
+ //Import Reaction File
+ BioCollection<BioReaction> input = new BioCollection<>();
+ try {
+ BioNetwork finalNetwork = network;
+ Mapper map = new Mapper(finalNetwork, bioNetwork -> {
+ return finalNetwork.getReactionsView();
+ })
+ .columnSeparator(sep)
+ .idColumn(i)
+ .skipIfNotFound();
+ if(hasHeader) map = map.skipHeader();
+ input = map.map(reactionFile);
+ System.err.println(input.size()+" reactions mapped");
+ System.err.println(map.getNumberOfSkippedEntries()+" reactions not found in model");
+ } catch (IOException e) {
+ System.err.println("Error while reading the Reaction file");
+ System.err.println(e.getMessage());
+ System.exit(1);
+ }
+
+ //default case: everything printed
+ if(!(printSubstrates|printProducts)){
+ printSubstrates=printProducts=true;
+ }
+
+ //Print output
+ try (PrintWriter writer = new PrintWriter(new FileWriter(this.outputFile, false))) {
+ for (BioReaction r : input){
+ BioCollection<BioMetabolite> metabolites = new BioCollection<>();
+ if(printSubstrates || r.isReversible()) metabolites.addAll(network.getLefts(r));
+ if(printProducts || r.isReversible()) metabolites.addAll(network.getRights(r));
+ for (BioMetabolite m : metabolites) {
+ writer.println(r.getId()+"\t"+m.getId());
+ }
+ }
+ } catch (IOException e) {
+ System.err.println("Error while writing SBML entities");
+ System.exit(1);
+ }
+ }
+
+ @Override
+ public String getLabel() {
+ return this.getClass().getSimpleName();
+ }
+
+ @Override
+ public String getLongDescription() {
+ return "Get reactants lists from a list of reactions and a GSMN. Output a tab-separated file " +
+ "with one row per reactant, reaction identifiers in first column, reactant identifiers in second column. " +
+ "It can provides substrates, products, or both (by default). In the case of reversible reactions, " +
+ "all reactants are considered both substrates and products";
+ }
+
+ @Override
+ public String getShortDescription() {
+ return "Get reactants lists from a list of reactions and a GSMN.";
+ }
+}
--
GitLab
From 3df40b4bd07cec01727d2294379890ca5a45cd4c Mon Sep 17 00:00:00 2001
From: cfrainay <clement.frainay@inrae.fr>
Date: Thu, 28 Jul 2022 18:10:04 +0200
Subject: [PATCH 3/3] Add app to get gene list from a list of reactions and
SBML GPR annotations
---
.../attributes/GetGenesFromReactions.java | 128 ++++++++++++++++++
1 file changed, 128 insertions(+)
create mode 100644 met4j-toolbox/src/main/java/fr/inrae/toulouse/metexplore/met4j_toolbox/attributes/GetGenesFromReactions.java
diff --git a/met4j-toolbox/src/main/java/fr/inrae/toulouse/metexplore/met4j_toolbox/attributes/GetGenesFromReactions.java b/met4j-toolbox/src/main/java/fr/inrae/toulouse/metexplore/met4j_toolbox/attributes/GetGenesFromReactions.java
new file mode 100644
index 000000000..777cba704
--- /dev/null
+++ b/met4j-toolbox/src/main/java/fr/inrae/toulouse/metexplore/met4j_toolbox/attributes/GetGenesFromReactions.java
@@ -0,0 +1,128 @@
+package fr.inrae.toulouse.metexplore.met4j_toolbox.attributes;
+
+import fr.inrae.toulouse.metexplore.met4j_core.biodata.BioGene;
+import fr.inrae.toulouse.metexplore.met4j_core.biodata.BioMetabolite;
+import fr.inrae.toulouse.metexplore.met4j_core.biodata.BioNetwork;
+import fr.inrae.toulouse.metexplore.met4j_core.biodata.BioReaction;
+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_mapping.Mapper;
+import fr.inrae.toulouse.metexplore.met4j_toolbox.generic.AbstractMet4jApplication;
+import fr.inrae.toulouse.metexplore.met4j_toolbox.generic.annotations.EnumFormats;
+import fr.inrae.toulouse.metexplore.met4j_toolbox.generic.annotations.EnumParameterTypes;
+import fr.inrae.toulouse.metexplore.met4j_toolbox.generic.annotations.Format;
+import fr.inrae.toulouse.metexplore.met4j_toolbox.generic.annotations.ParameterType;
+import org.kohsuke.args4j.Option;
+
+import java.io.FileWriter;
+import java.io.IOException;
+import java.io.PrintWriter;
+
+public class GetGenesFromReactions extends AbstractMet4jApplication {
+
+
+ @ParameterType(name= EnumParameterTypes.InputFile)
+ @Format(name= EnumFormats.Sbml)
+ @Option(name = "-i", usage = "Input SBML file", required = true)
+ public String sbml;
+
+ @ParameterType(name= EnumParameterTypes.InputFile)
+ @Format(name= EnumFormats.Tsv)
+ @Option(name = "-r", usage = "Input Reaction file", required = true)
+ public String reactionFile;
+
+ @ParameterType(name= EnumParameterTypes.Text)
+ @Option(name = "-sep", usage = "Separator in reaction file", required = false)
+ public String sep = "\t";
+ @ParameterType(name= EnumParameterTypes.Boolean)
+ @Option(name = "-header", usage = "Skip reaction file header", required = false)
+ public boolean hasHeader = false;
+ @ParameterType(name= EnumParameterTypes.Integer)
+ @Option(name = "-col", usage = "Column number in reaction file (first as 1)", required = false)
+ public int i=1;
+
+ @ParameterType(name= EnumParameterTypes.OutputFile)
+ @Format(name= EnumFormats.Tsv)
+ @Option(name = "-o", usage = "Output file", required=true)
+ public String outputFile;
+
+ /**
+ * <p>main.</p>
+ *
+ * @param args an array of {@link String} objects.
+ */
+ public static void main(String[] args) {
+ GetGenesFromReactions app = new GetGenesFromReactions();
+ app.parseArguments(args);
+ app.run();
+ }
+
+
+ /**
+ * <p>run.</p>
+ */
+ public void run() {
+
+ //read SBML, create bionetwork
+ String fileIn = this.sbml;
+ JsbmlReader reader = new JsbmlReader(fileIn);
+ BioNetwork network = null;
+ try {
+ network = reader.read();
+ } catch (Met4jSbmlReaderException e) {
+ System.err.println("Error while reading the SBML file");
+ System.err.println(e.getMessage());
+ System.exit(1);
+ }
+
+ //Import Reaction File
+ BioCollection<BioReaction> input = new BioCollection<>();
+ try {
+ BioNetwork finalNetwork = network;
+ Mapper map = new Mapper(finalNetwork, bioNetwork -> {
+ return finalNetwork.getReactionsView();
+ })
+ .columnSeparator(sep)
+ .idColumn(i)
+ .skipIfNotFound();
+ if(hasHeader) map = map.skipHeader();
+ input = map.map(reactionFile);
+ System.err.println(input.size()+" reactions mapped");
+ System.err.println(map.getNumberOfSkippedEntries()+" reactions not found in model");
+ } catch (IOException e) {
+ System.err.println("Error while reading the Reaction file");
+ System.err.println(e.getMessage());
+ System.exit(1);
+ }
+
+ //Print output
+ try (PrintWriter writer = new PrintWriter(new FileWriter(this.outputFile, false))) {
+ for (BioReaction r : input){
+ BioCollection<BioGene> genes = network.getGenesFromReactions(r);
+ for (BioGene g : genes) {
+ writer.println(r.getId()+"\t"+g.getId());
+ }
+ }
+ } catch (IOException e) {
+ System.err.println("Error while writing SBML entities");
+ System.exit(1);
+ }
+ }
+
+ @Override
+ public String getLabel() {
+ return this.getClass().getSimpleName();
+ }
+
+ @Override
+ public String getLongDescription() {
+ return "Get associated gene list from a list of reactions and a GSMN. Parse GSMN GPR annotations and output a tab-separated file " +
+ "with one row per gene, associated reaction identifiers from input file in first column, gene identifiers in second column.";
+ }
+
+ @Override
+ public String getShortDescription() {
+ return "Get gene lists from a list of reactions and a GSMN.";
+ }
+}
--
GitLab