Merge branch 'p2rank-2-4' into develop

rdk · Sep 9, 2024 · e39ffe5 · e39ffe5
2 parents d1cade7 + 1101acc
commit e39ffe5
Show file tree

Hide file tree

Showing 7 changed files with 78 additions and 16 deletions.
diff --git a/src/main/groovy/cz/siret/prank/domain/Dataset.groovy b/src/main/groovy/cz/siret/prank/domain/Dataset.groovy
@@ -68,6 +68,7 @@ class Dataset implements Parametrized, Writable, Failable {
     static final String COLUMN_CONSERVATION_FILES_PATTERN = "conservation_files_pattern"
     static final String COLUMN_APO_PROTEIN = "apo_protein"
     static final String COLUMN_APO_CHAINS = "apo_chains"
+    static final String COLUMN_POSITIVE_RESIDUES = "positive_residues"
 
     static final List<String> DEFAULT_HEADER = [ COLUMN_PROTEIN ]
 
@@ -423,7 +424,7 @@ class Dataset implements Parametrized, Writable, Failable {
      * @return true if explicit residue labeling is defined a as a part of th dataset
      */
     boolean hasExplicitResidueLabeling() {
-        return attributes.containsKey(PARAM_RESIDUE_LABELING_FORMAT)
+        return attributes.containsKey(PARAM_RESIDUE_LABELING_FORMAT) || header.contains(COLUMN_POSITIVE_RESIDUES)
     }
 
     /**
@@ -438,9 +439,16 @@ class Dataset implements Parametrized, Writable, Failable {
      */
     @Nullable
     ResidueLabeler getResidueLabeler() {
+
         if (residueLabeler == null && hasExplicitResidueLabeling()) {
-            String labelingFile = dir + "/" + attributes.get(PARAM_RESIDUE_LABELING_FILE)
-            residueLabeler = ResidueLabeler.loadFromFile(attributes.get(PARAM_RESIDUE_LABELING_FORMAT), labelingFile)
+
+            if (header.contains(COLUMN_POSITIVE_RESIDUES)) {
+
+            } else {
+                String labelingFile = dir + "/" + attributes.get(PARAM_RESIDUE_LABELING_FILE)
+                residueLabeler = ResidueLabeler.loadFromFile(attributes.get(PARAM_RESIDUE_LABELING_FORMAT), labelingFile)
+            }
+
         }
         return  residueLabeler
     }
@@ -883,7 +891,7 @@ class Dataset implements Parametrized, Writable, Failable {
         @Nullable
         BinaryLabeling getExplicitBinaryLabeling() {
             if (originDataset.hasExplicitResidueLabeling()) {
-                return originDataset.explicitBinaryResidueLabeler.getBinaryLabeling(protein.residues, protein)
+                return originDataset.explicitBinaryResidueLabeler.getBinaryLabeling(protein.residues, protein, this)
             }
             return null
         }
@@ -893,7 +901,7 @@ class Dataset implements Parametrized, Writable, Failable {
          */
         @Nullable
         BinaryLabeling getBinaryLabeling() {
-            return originDataset.binaryResidueLabeler.getBinaryLabeling(protein.residues, protein)
+            return originDataset.binaryResidueLabeler.getBinaryLabeling(protein.residues, protein, this)
         }
 
         ProcessedItemContext getContext() {

diff --git a/src/main/groovy/cz/siret/prank/domain/labeling/ColumnBasedResidueLabeler.groovy b/src/main/groovy/cz/siret/prank/domain/labeling/ColumnBasedResidueLabeler.groovy
@@ -0,0 +1,50 @@
+package cz.siret.prank.domain.labeling
+
+import cz.siret.prank.domain.Dataset
+import cz.siret.prank.domain.Protein
+import cz.siret.prank.domain.Residue
+import cz.siret.prank.domain.Residues
+import cz.siret.prank.utils.Sutils
+import cz.siret.prank.utils.Writable
+import groovy.transform.CompileStatic
+import groovy.util.logging.Slf4j
+
+/**
+ *
+ */
+@Slf4j
+@CompileStatic
+class ColumnBasedResidueLabeler extends ResidueLabeler<Boolean> implements Writable {
+
+    @Override
+    ResidueLabeling<Boolean> labelResidues(Residues residues, Protein protein, Dataset.Item item) {
+
+        Set<String> positives = Sutils.split(item.columnValues.get(Dataset.COLUMN_POSITIVE_RESIDUES), ',').toSet()
+
+        return createLabelingFromPositiveResidueCodes(residues, positives)
+    }
+
+    @Override
+    boolean isBinary() {
+        return true
+    }
+
+    @Override
+    ResidueLabeling<Double> getDoubleLabeling() {
+        return null
+    }
+
+//===========================================================================================================//
+
+    static ResidueLabeling<Boolean> createLabelingFromPositiveResidueCodes(Residues residues, Set<String> positives) {
+        ResidueLabeling<Boolean> labeling = new ResidueLabeling<Boolean>()
+
+        for (Residue residue : residues) {
+            boolean label = positives.contains(residue.toString())
+            labeling.add(residue, label)
+        }
+
+        return labeling
+    }
+
+}
diff --git a/src/main/groovy/cz/siret/prank/domain/labeling/LigandBasedResidueLabeler.groovy b/src/main/groovy/cz/siret/prank/domain/labeling/LigandBasedResidueLabeler.groovy
@@ -1,6 +1,6 @@
 package cz.siret.prank.domain.labeling
 
-
+import cz.siret.prank.domain.Dataset
 import cz.siret.prank.domain.Protein
 import cz.siret.prank.domain.Residue
 import cz.siret.prank.domain.Residues
@@ -22,7 +22,7 @@ class LigandBasedResidueLabeler extends ResidueLabeler<Boolean> implements Param
     ResidueLabeling<Double> lastLigandDistanceLabeling
 
     @Override
-    ResidueLabeling<Boolean> labelResidues(Residues residues, Protein protein) {
+    ResidueLabeling<Boolean> labelResidues(Residues residues, Protein protein, Dataset.Item item) {
 
         ResidueLabeling<Double> ligandDistanceLabeling = new ResidueLabeling<>(residues.count)
         for (Residue res : residues) {

diff --git a/src/main/groovy/cz/siret/prank/domain/labeling/ModelBasedResidueLabeler.groovy b/src/main/groovy/cz/siret/prank/domain/labeling/ModelBasedResidueLabeler.groovy
@@ -1,5 +1,6 @@
 package cz.siret.prank.domain.labeling
 
+import cz.siret.prank.domain.Dataset
 import cz.siret.prank.domain.Protein
 import cz.siret.prank.domain.Residue
 import cz.siret.prank.domain.Residues
@@ -83,7 +84,7 @@ class ModelBasedResidueLabeler extends ResidueLabeler<Boolean> implements Parame
     }
 
     @Override
-    ResidueLabeling<Boolean> labelResidues(Residues residues, Protein protein) {
+    ResidueLabeling<Boolean> labelResidues(Residues residues, Protein protein, Dataset.Item item) {
 
         ModelBasedPointLabeler predictor = new ModelBasedPointLabeler(model, context).withObserved(observedPoints)
 

diff --git a/src/main/groovy/cz/siret/prank/domain/labeling/ResidueLabeler.groovy b/src/main/groovy/cz/siret/prank/domain/labeling/ResidueLabeler.groovy
@@ -1,6 +1,6 @@
 package cz.siret.prank.domain.labeling
 
-
+import cz.siret.prank.domain.Dataset
 import cz.siret.prank.domain.Protein
 import cz.siret.prank.domain.Residues
 import cz.siret.prank.program.PrankException
@@ -14,23 +14,27 @@ import javax.annotation.Nullable
 @CompileStatic
 abstract class ResidueLabeler<L> {
 
-    abstract ResidueLabeling<L> labelResidues(Residues residues, Protein protein)
+    abstract ResidueLabeling<L> labelResidues(Residues residues, Protein protein, Dataset.Item item)
 
     abstract boolean isBinary()
 
     @Nullable
     abstract ResidueLabeling<Double> getDoubleLabeling()
 
-    BinaryLabeling getBinaryLabeling(Residues residues, Protein protein) {
+    BinaryLabeling getBinaryLabeling(Residues residues, Protein protein, @Nullable Dataset.Item item) {
         if (isBinary()) {
-            (BinaryLabeling) labelResidues(residues, protein)
+            (BinaryLabeling) labelResidues(residues, protein, item)
         } else {
             throw new PrankException("Residue labeler not binary!")
         }
     }
 
+    BinaryLabeling getBinaryLabeling(Residues residues, Protein protein) {
+        return getBinaryLabeling(residues, protein, null)
+    }
+
     BinaryLabeling getBinaryLabeling(Protein protein) {
-        getBinaryLabeling(protein.residues, protein)
+        getBinaryLabeling(protein.residues, protein, null)
     }
 
     static ResidueLabeler loadFromFile(String format, String fname) {

diff --git a/src/main/groovy/cz/siret/prank/domain/labeling/SprintLabelingLoader.groovy b/src/main/groovy/cz/siret/prank/domain/labeling/SprintLabelingLoader.groovy
@@ -47,12 +47,11 @@ class SprintLabelingLoader extends ResidueLabeler<Boolean> implements Writable {
     }
 
     @Override
-    ResidueLabeling<Boolean> labelResidues(Residues residues, Protein protein) {
+    ResidueLabeling<Boolean> labelResidues(Residues residues, Protein protein, Dataset.Item item) {
 
         Map<Residue.Key, Boolean> labelMap = new HashMap<>()
 
 
-
         boolean foundOneChain = false
         for (ResidueChain chain : protein.residueChains) {
             String chainCode = toElementCode(protein, chain)

diff --git a/src/main/groovy/cz/siret/prank/domain/labeling/StaticResidueLabeler.groovy b/src/main/groovy/cz/siret/prank/domain/labeling/StaticResidueLabeler.groovy
@@ -54,7 +54,7 @@ abstract class StaticResidueLabeler extends ResidueLabeler<Boolean> implements W
         }
 
         @Override
-        ResidueLabeling<Boolean> labelResidues(Residues residues, Protein protein) {
+        ResidueLabeling<Boolean> labelResidues(Residues residues, Protein protein, Dataset.Item item) {
 
             Map<String, Double> scores = new HashMap<>()
             for (String line : Futils.readLines(path).tail()) {