Merge branch 'develop' of github.com:PDB-REDO/libcifpp into develop

include/cif++/model.hpp

@@ -321,13 +321,6 @@ class atom
 
 	friend std::ostream &operator<<(std::ostream &os, const atom &atom);
 
-	// /// \brief Synchronize data with underlying cif data
-	// void sync()
-	// {
-	// 	if (m_impl)
-	// 		m_impl->prefetch();
-	// }
-
   private:
 	friend class structure;
 
@@ -341,25 +334,6 @@ class atom
 	std::shared_ptr<atom_impl> m_impl;
 };
 
-// template <>
-// inline std::string atom::get_property<std::string>(const std::string_view name) const
-// {
-// 	return get_property(name);
-// }
-
-// template <>
-// inline int atom::get_property<int>(const std::string_view name) const
-// {
-// 	auto v = impl().get_property(name);
-// 	return v.empty() ? 0 : stoi(v);
-// }
-
-// template <>
-// inline float atom::get_property<float>(const std::string_view name) const
-// {
-// 	return stof(impl().get_property(name));
-// }
-
 inline void swap(atom &a, atom &b)
 {
 	a.swap(b);
@@ -949,4 +923,11 @@ class structure
 	std::vector<residue> m_non_polymers;
 };
 
+// --------------------------------------------------------------------
+
+/// \brief Reconstruct all missing categories for an assumed PDBx file.
+/// Some people believe that simply dumping some atom records is enough.
+/// \param db The cif::datablock that hopefully contains some valid data
+void reconstruct_pdbx(datablock &db);
+
 } // namespace cif::mm

src/model.cpp

@@ -618,19 +618,19 @@ float monomer::phi() const
 {
 	float result = 360;
 
-	try
-	{
 	if (m_index > 0)
 	{
 		auto &prev = m_polymer->operator[](m_index - 1);
 		if (prev.m_seq_id + 1 == m_seq_id)
-				result = static_cast<float>(dihedral_angle(prev.C().get_location(), N().get_location(), CAlpha().get_location(), C().get_location()));
-		}
-	}
-	catch (const std::exception &ex)
 		{
-		if (VERBOSE > 0)
-			std::cerr << ex.what() << std::endl;
+			auto a1 = prev.C();
+			auto a2 = N();
+			auto a3 = CAlpha();
+			auto a4 = C();
+
+			if (a1 and a2 and a3 and a4)
+				result = dihedral_angle(a1.get_location(), a2.get_location(), a3.get_location(), a4.get_location());
+		}
 	}
 
 	return result;
@@ -640,19 +640,19 @@ float monomer::psi() const
 {
 	float result = 360;
 
-	try
-	{
 	if (m_index + 1 < m_polymer->size())
 	{
 		auto &next = m_polymer->operator[](m_index + 1);
 		if (m_seq_id + 1 == next.m_seq_id)
-				result = static_cast<float>(dihedral_angle(N().get_location(), CAlpha().get_location(), C().get_location(), next.N().get_location()));
-		}
-	}
-	catch (const std::exception &ex)
 		{
-		if (VERBOSE > 0)
-			std::cerr << ex.what() << std::endl;
+			auto a1 = N();
+			auto a2 = CAlpha();
+			auto a3 = C();
+			auto a4 = next.N();
+
+			if (a1 and a2 and a3 and a4)
+				result = dihedral_angle(a1.get_location(), a2.get_location(), a3.get_location(), a4.get_location());
+		}
 	}
 
 	return result;
@@ -935,17 +935,17 @@ float monomer::omega(const monomer &a, const monomer &b)
 {
 	float result = 360;
 
-	try
-	{
+	auto a1 = a.get_atom_by_atom_id("CA");
+	auto a2 = a.get_atom_by_atom_id("C");
+	auto a3 = b.get_atom_by_atom_id("N");
+	auto a4 = b.get_atom_by_atom_id("CA");
+
+	if (a1 and a2 and a3 and a4)
 		result = static_cast<float>(dihedral_angle(
-			a.get_atom_by_atom_id("CA").get_location(),
-			a.get_atom_by_atom_id("C").get_location(),
-			b.get_atom_by_atom_id("N").get_location(),
-			b.get_atom_by_atom_id("CA").get_location()));
-	}
-	catch (...)
-	{
-	}
+			a1.get_location(),
+			a2.get_location(),
+			a3.get_location(),
+			a4.get_location()));
 
 	return result;
 }
@@ -974,8 +974,12 @@ polymer::polymer(structure &s, const std::string &entityID, const std::string &a
 	for (auto r : poly_seq_scheme.find("asym_id"_key == asym_id))
 	{
 		int seqID;
+		std::optional<int> pdbSeqNum;
 		std::string compoundID, authSeqID, pdbInsCode;
-		cif::tie(seqID, authSeqID, compoundID, pdbInsCode) = r.get("seq_id", "auth_seq_num", "mon_id", "pdb_ins_code");
+		cif::tie(seqID, authSeqID, compoundID, pdbInsCode, pdbSeqNum) = r.get("seq_id", "auth_seq_num", "mon_id", "pdb_ins_code", "pdb_seq_num");
+
+		if (authSeqID.empty() and pdbSeqNum.has_value())
+			authSeqID = std::to_string(*pdbSeqNum);
 
 		size_t index = size();
 
@@ -2773,4 +2777,14 @@ void structure::validate_atoms() const
 	assert(atoms.empty());
 }
 
+// --------------------------------------------------------------------
+
+void reconstruct_pdbx(datablock &db)
+{
+	if (db.get("atom_site") == nullptr)
+		throw std::runtime_error("Cannot reconstruct PDBx file, atom data missing");
+	
+	
+}
+
 } // namespace pdbx

src/pdb/pdb2cif.cpp

@@ -3819,27 +3819,33 @@ void PDBFileParser::ConstructEntities()
 
 			for (std::string monID : monIds)
 			{
-				std::string authMonID, authSeqNum, authInsCode;
+				std::string authMonID, authSeqNum, authInsCode{'.'};
+
 				if (res.mSeen)
 				{
 					authMonID = monID;
 					authSeqNum = std::to_string(res.mSeqNum);
 					if (res.mIcode != ' ' and res.mIcode != 0)
 						authInsCode = std::string{ res.mIcode };
+
+					cat->emplace({
+						{ "asym_id", asymID },
+						{ "entity_id", mMolID2EntityID[chain.mMolID] },
+						{ "seq_id", seqID },
+						{ "mon_id", monID },
+						{ "ndb_seq_num", seqID },
+						{ "pdb_seq_num", res.mSeqNum },
+						{ "auth_seq_num", authSeqNum },
+						{ "pdb_mon_id", authMonID },
+						{ "auth_mon_id", authMonID },
+						{ "pdb_strand_id", std::string{ chain.mDbref.chainID } },
+						{ "pdb_ins_code", authInsCode },
+						{ "hetero", res.mAlts.empty() ? "n" : "y" } });
 				}
 				else
 				{
-					authMonID = res.mMonID;
-					authSeqNum = std::to_string(res.mSeqNum);
-
 					if (res.mIcode != ' ' and res.mIcode != 0)
 						authInsCode = std::string{ res.mIcode } + "A";
-					else
-						authInsCode = "A";
-				}
-
-				if (authInsCode.empty())
-					authInsCode = ".";
 
 					cat->emplace({
 						{ "asym_id", asymID },
@@ -3848,15 +3854,16 @@ void PDBFileParser::ConstructEntities()
 						{ "mon_id", monID },
 						{ "ndb_seq_num", seqID },
 						{ "pdb_seq_num", res.mSeqNum },
-					{ "auth_seq_num", authSeqNum },
-					{ "pdb_mon_id", authMonID },
-					{ "auth_mon_id", authMonID },
+						{ "auth_seq_num", "." },
+						{ "pdb_mon_id", "." },
+						{ "auth_mon_id", "." },
 						{ "pdb_strand_id", std::string{ chain.mDbref.chainID } },
 						{ "pdb_ins_code", authInsCode },
 						{ "hetero", res.mAlts.empty() ? "n" : "y" } });
 				}
 			}
 		}
+	}
 
 	// We have now created all compounds, write them out
 	uint32_t structRefID = 0, structRefSeqAlignID = 0;