fix discontinued chains 2

rdk · Apr 11, 2024 · ca5670f · ca5670f
1 parent 2035ca9
commit ca5670f
Show file tree

Hide file tree

Showing 6 changed files with 10,274 additions and 17 deletions.
diff --git a/build.gradle b/build.gradle
@@ -18,7 +18,7 @@ apply plugin: 'groovy'
 apply plugin: 'java'
 
 group = 'cz.siret'
-version = '2.5.0-dev.4.2'
+version = '2.5.0-dev.4.3'
 
 
 description = 'Ligand binding site prediction based on machine learning.'

diff --git a/src/main/groovy/cz/siret/prank/geom/Struct.groovy b/src/main/groovy/cz/siret/prank/geom/Struct.groovy
@@ -217,8 +217,8 @@ class Struct {
     }
 
     private static int getLastTerminalResidue(List<Group> chainGroups) {
-        for (int i=chainGroups.size()-1; i<=0; i--) {
-            if (isTerminalResidue(chainGroups[i])) {
+        for (int i=chainGroups.size()-1; i>=0; i--) {
+            if (isAminoAcidResidueHeuristic(i, chainGroups) && isTerminalResidue(chainGroups[i])) {
                 return i
             }
         }
@@ -231,17 +231,22 @@ class Struct {
         int n = chainGroups.size()
         List<Group> res = new ArrayList<>(n)
 
-        log.info "groups in chain {}: {}", getAuthorId(chain), n
+        log.info "all groups in chain {}: {}", getAuthorId(chain), n
 
-        int lastTerminalResidueIdx = getLastTerminalResidue(chainGroups)
+        //int lastTerminalResidueIdx = getLastTerminalResidue(chainGroups)
+        //log.info "lastTerminalResidueIdx: {}", lastTerminalResidueIdx
 
         for (int i=0; i!=n; i++) {
+            Group g = chainGroups[i]
             if (isAminoAcidResidueHeuristic(i, chainGroups)) {
-                Group g = chainGroups[i]
                 res.add g
-                if (i == lastTerminalResidueIdx) {
-                    break // this is done so amino acid ligands at the end are excluded
-                }
+
+                // Note: commented out because this just doesn't work in proteins with discontinued chains (OXT atom is not present at the last residue)
+                //if (i == lastTerminalResidueIdx) {
+                //    break // this is done so amino acid ligands at the end are excluded
+                //}
+            } else {
+                log.warn "group {} ({}) considered non-protein", i, g.getPDBName()
             }
         }
 

diff --git a/src/test/groovy/cz/siret/prank/DiscontinuedChainsTest.groovy b/src/test/groovy/cz/siret/prank/DiscontinuedChainsTest.groovy
@@ -14,8 +14,11 @@ import static org.junit.jupiter.api.Assertions.assertEquals
 @CompileStatic
 class DiscontinuedChainsTest {
 
-    String PDB_1O6U = 'src/test/resources/data/tricky_cases/1o6u.pdb'
-    String CIF_1O6U = 'src/test/resources/data/tricky_cases/1o6u.cif'
+    String DIR = 'src/test/resources/data/tricky_cases/'
+    String PDB_1O6U = DIR +'1o6u.pdb'
+    String CIF_1O6U = DIR +'1o6u.cif'
+    String PDB_1UWY = DIR +'1uwy.pdb'
+    String CIF_1UWY = DIR +'1uwy.cif'
 
     @Test
     void discontinuedChainsLoading_1O6U_pdb() throws Exception {
@@ -57,11 +60,44 @@ class DiscontinuedChainsTest {
         def chainA = prot.getResidueChain('A')
         def chainC = prot.getResidueChain('C')
 
-        assertEquals(394, chainA.length, "Expected length of chain A")
+        assertEquals(394, chainA.length, "Expected length of chain A")  // TODO pdb says different
         assertEquals(387, chainC.length, "Expected length of chain C")
         assertEquals(10203, prot.allAtoms.count, "Expected number of all structure atoms")
         assertEquals(9311, prot.proteinAtoms.count, "Expected number of protein atoms")
 
     }
+
+
+    @Test
+    void discontinuedChainsLoading_1UWY_pdb() throws Exception {
+        Protein prot = Protein.load(PDB_1UWY)
+
+        def chainA = prot.getResidueChain('A')
+
+        log.info "All Atoms: {}", prot.allAtoms.count
+        log.info "Protein Atoms: {}", prot.proteinAtoms.count
+        log.info "Chain A Length: {}", chainA.length
+
+        assertEquals(394, chainA.length, "Expected length of chain A") // TODO pdb says 426!
+        assertEquals(3230, prot.allAtoms.count, "Expected number of all structure atoms")
+        assertEquals(3116, prot.proteinAtoms.count, "Expected number of protein atoms")
+
+    }
+
+    @Test
+    void discontinuedChainsLoading_1UWY_cif() throws Exception {
+        Protein prot = Protein.load(CIF_1UWY)
+
+        def chainA = prot.getResidueChain('A')
+
+        log.info "All Atoms: {}", prot.allAtoms.count
+        log.info "Protein Atoms: {}", prot.proteinAtoms.count
+        log.info "Chain A Length: {}", chainA.length
+
+        assertEquals(394, chainA.length, "Expected length of chain A") // TODO pdb says 426!
+        assertEquals(3230, prot.allAtoms.count, "Expected number of all structure atoms")
+        assertEquals(3116, prot.proteinAtoms.count, "Expected number of protein atoms")
+
+    }
 
 }