change getResidue

CMakeLists.txt

@@ -25,7 +25,7 @@
 cmake_minimum_required(VERSION 3.16)
 
 # set the project name
-project(cifpp VERSION 3.0.5 LANGUAGES CXX)
+project(cifpp VERSION 4.0.0 LANGUAGES CXX)
 
 list(PREPEND CMAKE_MODULE_PATH "${CMAKE_CURRENT_SOURCE_DIR}/cmake")
 

changelog

+Version 4.0.0
+- getResidue in mmcif::Structure now requires both a
+  sequence ID and an auth sequence ID. Very unfortunate.
+
 Version 3.0.5
 - mmcif::Structure redesign. It is now a wrapper around a cif::Datablock.
 

include/cif++/Structure.hpp

@@ -296,7 +296,7 @@ class Residue
   public:
 	// constructor
 	Residue(const Structure &structure, const std::string &compoundID,
-		const std::string &asymID, int seqID = 0, const std::string &authSeqID = {})
+		const std::string &asymID, int seqID, const std::string &authSeqID)
 		: mStructure(&structure)
 		, mCompoundID(compoundID)
 		, mAsymID(asymID)
@@ -383,9 +383,6 @@ class Residue
 	const Structure *mStructure = nullptr;
 	std::string mCompoundID, mAsymID;
 	int mSeqID = 0;
-
-	// Watch out, this is used only to label waters... The rest of the code relies on
-	// MapLabelToAuth to get this info. Perhaps we should rename this member field.
 	std::string mAuthSeqID;
 	AtomView mAtoms;
 };
@@ -625,26 +622,41 @@ class Structure
 	/// \brief Return the atom closest to point \a p with atom type \a type in a residue of type \a res_type
 	Atom getAtomByPositionAndType(Point p, std::string_view type, std::string_view res_type) const;
 
-	/// \brief Get a residue, if \a seqID is zero, the non-polymers are searched
-	const Residue &getResidue(const std::string &asymID, const std::string &compID, int seqID = 0) const;
+	/// \brief Get a residue for an asym with id \a asymID seq id \a seqID and authSeqID \a authSeqID
+	Residue &getResidue(const std::string &asymID, int seqID, const std::string &authSeqID);
 
-	/// \brief Get a residue, if \a seqID is zero, the non-polymers are searched
-	Residue &getResidue(const std::string &asymID, const std::string &compID, int seqID = 0);
+	/// \brief Get a the single residue for an asym with id \a asymID seq id \a seqID and authSeqID \a authSeqID
+	const Residue &getResidue(const std::string &asymID, int seqID, const std::string &authSeqID) const
+	{
+		return const_cast<Structure*>(this)->getResidue(asymID, seqID, authSeqID);
+	}
 
-	/// \brief Get a the single residue for an asym with id \a asymID
-	const Residue &getResidue(const std::string &asymID) const;
+	/// \brief Get a residue for an asym with id \a asymID, compound id \a compID, seq id \a seqID and authSeqID \a authSeqID
+	Residue &getResidue(const std::string &asymID, const std::string &compID, int seqID, const std::string &authSeqID);
 
-	/// \brief Get a the single residue for an asym with id \a asymID and seq id \a seqID
-	const Residue &getResidue(const std::string &asymID, int seqID) const;
+	/// \brief Get a residue for an asym with id \a asymID, compound id \a compID, seq id \a seqID and authSeqID \a authSeqID
+	const Residue &getResidue(const std::string &asymID, const std::string &compID, int seqID, const std::string &authSeqID) const
+	{
+		return const_cast<Structure*>(this)->getResidue(asymID, compID, seqID, authSeqID);
+	}
 
 	/// \brief Get a the single residue for an asym with id \a asymID
 	Residue &getResidue(const std::string &asymID);
 
+	/// \brief Get a the single residue for an asym with id \a asymID
+	const Residue &getResidue(const std::string &asymID) const
+	{
+		return const_cast<Structure*>(this)->getResidue(asymID);
+	}
+
 	/// \brief Get a the residue for atom \a atom
 	Residue &getResidue(const mmcif::Atom &atom);
 
 	/// \brief Get a the residue for atom \a atom
-	const Residue &getResidue(const mmcif::Atom &atom) const;
+	const Residue &getResidue(const mmcif::Atom &atom) const
+	{
+		return const_cast<Structure*>(this)->getResidue(atom);
+	}
 
 	// map between auth and label locations
 
@@ -697,11 +709,14 @@ class Structure
 	/// \brief Create a new (sugar) branch with one first NAG containing atoms \a nag_atoms
 	Branch& createBranch(std::vector<std::vector<cif::Item>> &nag_atoms);
 
+	/// \brief Extend an existing (sugar) branch identified by \a asymID with one sugar containing atoms \a atoms
+	Branch& extendBranch(const std::string &asym_id, std::vector<std::vector<cif::Item>> &atoms);
+
 	/// \brief Remove a residue, can be monomer or nonpoly
 	///
 	/// \param asym_id     The asym ID
 	/// \param seq_id      The sequence ID
-	void removeResidue(const std::string &sym_id, int seq_id);
+	void removeResidue(const std::string &asym_id, int seq_id);
 
 	/// \brief To sort the atoms in order of model > asym-id > res-id > atom-id
 	/// Will asssign new atom_id's to all atoms. Be carefull

src/BondMap.cpp

@@ -255,10 +255,10 @@ BondMap::BondMap(const Structure &p)
 	cif::Progress progress(compounds.size(), "Creating bond map");
 
 	// some helper indices to speed things up a bit
-	std::map<std::tuple<std::string, int, std::string>, std::string> atomMapByAsymSeqAndAtom;
+	std::map<std::tuple<std::string, int, std::string, std::string>, std::string> atomMapByAsymSeqAndAtom;
 	for (auto &a : p.atoms())
 	{
-		auto key = make_tuple(a.labelAsymID(), a.labelSeqID(), a.labelAtomID());
+		auto key = make_tuple(a.labelAsymID(), a.labelSeqID(), a.labelAtomID(), a.authSeqID());
 		atomMapByAsymSeqAndAtom[key] = a.id();
 	}
 
@@ -280,8 +280,8 @@ BondMap::BondMap(const Structure &p)
 			continue;
 		}
 
-		auto c = atomMapByAsymSeqAndAtom[make_tuple(asymID, lastSeqID, "C")];
-		auto n = atomMapByAsymSeqAndAtom[make_tuple(asymID, seqID, "N")];
+		auto c = atomMapByAsymSeqAndAtom[make_tuple(asymID, lastSeqID, "C", "")];
+		auto n = atomMapByAsymSeqAndAtom[make_tuple(asymID, seqID, "N", "")];
 
 		if (not(c.empty() or n.empty()))
 			bindAtoms(c, n);
@@ -293,14 +293,16 @@ BondMap::BondMap(const Structure &p)
 	{
 		std::string asym1, asym2, atomId1, atomId2;
 		int seqId1 = 0, seqId2 = 0;
-		cif::tie(asym1, asym2, atomId1, atomId2, seqId1, seqId2) =
+		std::string authSeqId1, authSeqId2;
+
+		cif::tie(asym1, asym2, atomId1, atomId2, seqId1, seqId2, authSeqId1, authSeqId2) =
 			l.get("ptnr1_label_asym_id", "ptnr2_label_asym_id",
 				"ptnr1_label_atom_id", "ptnr2_label_atom_id",
-				"ptnr1_label_seq_id", "ptnr2_label_seq_id");
-
-		std::string a = atomMapByAsymSeqAndAtom[make_tuple(asym1, seqId1, atomId1)];
-		std::string b = atomMapByAsymSeqAndAtom[make_tuple(asym2, seqId2, atomId2)];
+				"ptnr1_label_seq_id", "ptnr2_label_seq_id",
+				"ptnr1_auth_seq_id", "ptnr2_auth_seq_id");
 
+		std::string a = atomMapByAsymSeqAndAtom[make_tuple(asym1, seqId1, atomId1, authSeqId1)];
+		std::string b = atomMapByAsymSeqAndAtom[make_tuple(asym2, seqId2, atomId2, authSeqId2)];
 		if (not(a.empty() or b.empty()))
 			linkAtoms(a, b);
 	}
@@ -373,15 +375,12 @@ BondMap::BondMap(const Structure &p)
 		}
 
 		// loop over pdbx_branch_scheme
-		for (auto r : db["pdbx_branch_scheme"].find(cif::Key("mon_id") == c))
+		for (const auto &[asym_id, pdb_seq_num] : db["pdbx_branch_scheme"].find<std::string,std::string>(cif::Key("mon_id") == c, "asym_id", "pdb_seq_num"))
 		{
-			std::string asymID;
-			cif::tie(asymID) = r.get("asym_id");
-
 			std::vector<Atom> rAtoms;
 			copy_if(atoms.begin(), atoms.end(), back_inserter(rAtoms),
-				[&](auto &a)
-				{ return a.labelAsymID() == asymID; });
+				[&](const Atom &a)
+				{ return a.labelAsymID() == asym_id and a.authSeqID() == pdb_seq_num; });
 
 			for (uint32_t i = 0; i + 1 < rAtoms.size(); ++i)
 			{

test/rename-compound-test.cpp

@@ -32,7 +32,7 @@ int main(int argc, char* argv[])
 		mmcif::File f(testdir / ".."/"examples"/"1cbs.cif.gz");
 		mmcif::Structure structure(f);
 
-		auto &res = structure.getResidue("B", "REA");
+		auto &res = structure.getResidue("B");
 		structure.changeResidue(res, "RXA", {});
 
 		structure.cleanupEmptyCategories();

test/structure-test.cpp

@@ -123,8 +123,8 @@ _pdbx_nonpoly_scheme.asym_id         A
 _pdbx_nonpoly_scheme.ndb_seq_num     1 
 _pdbx_nonpoly_scheme.entity_id       1 
 _pdbx_nonpoly_scheme.mon_id          HEM 
-_pdbx_nonpoly_scheme.pdb_seq_num     0 
-_pdbx_nonpoly_scheme.auth_seq_num    0 
+_pdbx_nonpoly_scheme.pdb_seq_num     1 
+_pdbx_nonpoly_scheme.auth_seq_num    1 
 _pdbx_nonpoly_scheme.pdb_mon_id      HEM 
 _pdbx_nonpoly_scheme.auth_mon_id     HEM 
 _pdbx_nonpoly_scheme.pdb_strand_id   A 

test/unit-test.cpp

@@ -1634,7 +1634,7 @@ PRO OXT HXT SING N N 17
 
 	for (const auto &[compound, seqnr] : std::initializer_list<std::tuple<std::string, int>>{{"PRO", 1}, {"ASN", 2}, {"PHE", 3}})
 	{
-		auto &res = structure.getResidue("A", compound, seqnr);
+		auto &res = structure.getResidue("A", compound, seqnr, "");
 		auto atoms = res.atoms();
 
 		auto dc = components.get(compound);