added trivial mediator double bond position lookup

lifs-tools · Dec 1, 2023 · b1d2c26 · b1d2c26
1 parent f958589
commit b1d2c26
Show file tree

Hide file tree

Showing 3 changed files with 44 additions and 4 deletions.
diff --git a/parsers/pom.xml b/parsers/pom.xml
@@ -36,6 +36,7 @@
                                     <includes>
                                         <include>lipid-list.csv</include>
                                         <include>functional-groups.csv</include>
+                                        <include>trivial_mediators.csv</include>
                                         <include>**/*.g4</include>
                                     </includes>
                                 </resource>

diff --git a/parsers/src/main/java/org/lifstools/jgoslin/domain/KnownFunctionalGroups.java b/parsers/src/main/java/org/lifstools/jgoslin/domain/KnownFunctionalGroups.java
@@ -33,11 +33,25 @@
 public final class KnownFunctionalGroups extends HashMap<String, FunctionalGroup> {
 
     public static final int UNDEFINED_CLASS = 0;
+    private final HashMap<String, ArrayList<Integer>> trivialMediators = new HashMap<>();
 
-    private void loadData(List<String> lines, SumFormulaParser sumFormulaParser) {
+    private void loadData(List<String> lines, List<String> linesTM, SumFormulaParser sumFormulaParser) {
         int lineCounter = 0;
         ArrayList< ArrayList<String>> functional_data = new ArrayList<>();
         HashSet<String> functional_data_set = new HashSet<>();
+
+        for (String line : linesTM) {
+            if (line.isEmpty()) continue;
+
+            ArrayList<String> tokens = StringFunctions.splitString(line, '\t', '"', true);
+            if (tokens.size() != 2 || containsKey(tokens.get(0))) continue;
+
+            ArrayList<Integer> positions = new ArrayList<>();
+            for (String pos : StringFunctions.splitString(tokens.get(1), ',', '"', true)){
+                positions.add(Integer.parseInt(pos));
+            }
+            trivialMediators.put(tokens.get(0), positions);
+        }
 
         for (String line : lines) {
             if (lineCounter++ == 0) {
@@ -89,10 +103,18 @@ public KnownFunctionalGroups() {
 
     public KnownFunctionalGroups(List<String> lines, SumFormulaParser sumFormulaParser) {
         super();
-        loadData(lines, sumFormulaParser);
+        loadData(lines, StringFunctions.getResourceAsStringList(new ClassPathResource("trivial_mediators.csv")), sumFormulaParser);
     }
 
     public FunctionalGroup get(String s) {
         return super.get(s).copy();
     }
+
+    public HashMap<String, ArrayList<Integer>> getTmDb() {
+        return trivialMediators;
+    }
+
+    public ArrayList<Integer> db(String s) {
+        return trivialMediators.get(s);
+    }
 }
diff --git a/parsers/src/main/java/org/lifstools/jgoslin/parser/GoslinParserEventHandler.java b/parsers/src/main/java/org/lifstools/jgoslin/parser/GoslinParserEventHandler.java
@@ -36,6 +36,7 @@
 import static java.util.Map.entry;
 import java.util.TreeMap;
 import org.lifstools.jgoslin.domain.Element;
+import static org.lifstools.jgoslin.parser.Parser.EOF_SIGN;
 
 /**
  * Event handler implementation for the {@link GoslinParser}.
@@ -53,6 +54,7 @@ public class GoslinParserEventHandler extends LipidBaseParserEventHandler {
     private boolean mediatorSuffix;
     private Element heavyElement;
     private int heavyElementNumber;
+    private boolean trivialMediator;
 
     private final static Map<String, Integer> MEDIATOR_FA = Map.of(
             "H", 17, "O", 18, "E", 20, "Do", 22);
@@ -162,6 +164,7 @@ protected void resetParser(TreeNode node) {
         headgroupDecorators.clear();
         heavyElement = Element.C;
         heavyElementNumber = 0;
+        trivialMediator = false;
     }
 
     private void setHeadGroupName(TreeNode node) {
@@ -265,6 +268,19 @@ private void buildLipid(TreeNode node) {
         }
 
         Headgroup headgroup = prepareHeadgroupAndChecks();
+        String lipidName = node.getText();
+        if (lipidName.charAt(lipidName.length() - 1) == EOF_SIGN) lipidName = lipidName.substring(0, lipidName.length() - 1);
+        HashMap<String, ArrayList<Integer>> trivialDb = knownFunctionalGroups.getTmDb();
+
+        if (trivialMediator && trivialDb.containsKey(lipidName)){
+            ArrayList<Integer> dbPos = trivialDb.get(lipidName);
+            faList.get(0).getDoubleBonds().setNumDoubleBonds(dbPos.size());
+            Map<Integer, String> doubleBondPositions = faList.get(0).getDoubleBonds().getDoubleBondPositions();
+            doubleBondPositions.clear();
+
+            for (Integer p : dbPos) doubleBondPositions.put(p, "");
+            level = LipidLevel.FULL_STRUCTURE;
+        }
 
         LipidAdduct lipid = new LipidAdduct(assembleLipid(headgroup), adduct);
         content = lipid;
@@ -368,6 +384,7 @@ private void setMediatorTetranor(TreeNode node) {
     }
 
     private void setMediatorCarbon(TreeNode node) {
+        trivialMediator = true;
         currentFa.setNumCarbon(currentFa.getNumCarbon() + MEDIATOR_FA.get(node.getText()));
     }
 
@@ -398,13 +415,13 @@ private void addMediatorFunction(TreeNode node) {
             if (mediatorFunctionPositions.size() > 0) {
                 functionalGroup.setPosition(mediatorFunctionPositions.get(0));
             }
-        } else if (mediatorFunction.equals("Hp")) {
+        } else if (mediatorFunction.toLowerCase().equals("hp")) {
             functionalGroup = knownFunctionalGroups.get("OOH");
             fg = "OOH";
             if (mediatorFunctionPositions.size() > 0) {
                 functionalGroup.setPosition(mediatorFunctionPositions.get(0));
             }
-        } else if (mediatorFunction.equals("E") || mediatorFunction.equals("Ep")) {
+        } else if (mediatorFunction.equals("E") || mediatorFunction.toLowerCase().equals("ep")) {
             functionalGroup = knownFunctionalGroups.get("Ep");
             fg = "Ep";
             if (mediatorFunctionPositions.size() > 0) {