fix bug where some discontinued chains were not fully loaded as prote…

…in (e.g. in 1O6U)

build.gradle

@@ -18,7 +18,7 @@ apply plugin: 'groovy'
 apply plugin: 'java'
 
 group = 'cz.siret'
-version = '2.5.0-dev.4.1'
+version = '2.5.0-dev.4.2'
 
 
 description = 'Ligand binding site prediction based on machine learning.'

src/main/groovy/cz/siret/prank/geom/Struct.groovy

@@ -216,6 +216,15 @@ class Struct {
         group?.getAtom("OXT") != null
     }
 
+    private static int getLastTerminalResidue(List<Group> chainGroups) {
+        for (int i=chainGroups.size()-1; i<=0; i--) {
+            if (isTerminalResidue(chainGroups[i])) {
+                return i
+            }
+        }
+        return -1
+    }
+
     static List<Group> getResidueGroupsFromChain(Chain chain) {
 
         List<Group> chainGroups = chain.getAtomGroups()
@@ -224,11 +233,13 @@ class Struct {
 
         log.info "groups in chain {}: {}", getAuthorId(chain), n
 
+        int lastTerminalResidueIdx = getLastTerminalResidue(chainGroups)
+
         for (int i=0; i!=n; i++) {
             if (isAminoAcidResidueHeuristic(i, chainGroups)) {
                 Group g = chainGroups[i]
                 res.add g
-                if (isTerminalResidue(g)) {
+                if (i == lastTerminalResidueIdx) {
                     break // this is done so amino acid ligands at the end are excluded
                 }
             }

src/test/groovy/cz/siret/prank/DiscontinuedChainsTest.groovy

@@ -0,0 +1,67 @@
+package cz.siret.prank
+
+import cz.siret.prank.domain.Protein
+import groovy.transform.CompileStatic
+import groovy.util.logging.Slf4j
+import org.junit.jupiter.api.Test
+
+import static org.junit.jupiter.api.Assertions.assertEquals
+
+/**
+ *
+ */
+@Slf4j
+@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'
+
+    @Test
+    void discontinuedChainsLoading_1O6U_pdb() throws Exception {
+        Protein prot = Protein.load(PDB_1O6U)
+
+        def chainA = prot.getResidueChain('A')
+        def chainC = prot.getResidueChain('C')
+
+
+        log.info "All Atoms: {}", prot.allAtoms.count
+        log.info "Protein Atoms: {}", prot.proteinAtoms.count
+        log.info "Chain A Length: {}", chainA.length
+        log.info "Chain C Length: {}", chainC.length
+
+        assertEquals(394, chainA.length, "Expected length of chain A")
+        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")
+
+
+        // P2Rank 2.4.1
+        // All Atoms: 10203
+        // Protein Atoms: 8229
+        // Chain A Length: 321
+        // Chain C Length: 322
+
+        // P2Rank after fix
+        // All Atoms: 10203
+        // Protein Atoms: 9311
+        // Chain A Length: 394
+        // Chain C Length: 387
+
+    }
+
+    @Test
+    void discontinuedChainsLoading_1O6U_cif() throws Exception {
+        Protein prot = Protein.load(CIF_1O6U)
+
+        def chainA = prot.getResidueChain('A')
+        def chainC = prot.getResidueChain('C')
+
+        assertEquals(394, chainA.length, "Expected length of chain A")
+        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")
+
+    }
+
+}

src/test/groovy/cz/siret/prank/program/params/ConfigLoaderTest.groovy

@@ -24,4 +24,6 @@ class ConfigLoaderTest {
         assert p.seed == 42
     }
 
+    // TODO test rescore params
+
 }