Add formula_weight to entity in pdb2cif

src/pdb/pdb2cif.cpp

@@ -4511,6 +4511,47 @@ void PDBFileParser::ConstructEntities()
 			}
 			}
 		}
 		}
 	}
 	}
+
+	// Finish by calculating the formula_weight for each entity
+	for (auto entity : *getCategory("entity"))
+	{
+		auto entity_id = entity["id"].as<std::string>();
+		float formula_weight = 0;
+
+		if (entity["type"] == "polymer")
+		{
+			int n = 0;
+
+			for (std::string comp_id : getCategory("pdbx_poly_seq_scheme")->find<std::string>(cif::key("entity_id") == entity_id, "mon_id"))
+			{
+				auto compound = cif::compound_factory::instance().create(comp_id);
+				assert(compound);
+				if (not compound)
+					throw std::runtime_error("missing information for compound " + comp_id);
+				formula_weight += compound->formula_weight();
+				++n;
+			}
+
+			formula_weight -= (n - 1) * 18.015;
+		}
+		else if (entity["type"] == "water")
+			formula_weight = 18.015;
+		else
+		{
+			auto comp_id = getCategory("pdbx_nonpoly_scheme")->find_first<std::optional<std::string>>(cif::key("entity_id") == entity_id, "mon_id");
+			if (comp_id.has_value())
+			{
+				auto compound = cif::compound_factory::instance().create(*comp_id);
+				assert(compound);
+				if (not compound)
+					throw std::runtime_error("missing information for compound " + *comp_id);
+				formula_weight = compound->formula_weight();
+			}
+		}
+
+		if (formula_weight > 0)
+			entity["formula_weight"] = formula_weight;
+	}
 }
 }
 
 
 void PDBFileParser::ConstructSugarTrees(int &asymNr)
 void PDBFileParser::ConstructSugarTrees(int &asymNr)

test/unit-v2-test.cpp

@@ -3468,3 +3468,22 @@ TEST_CASE("compound_not_found_test_1")
 	auto cmp = cif::compound_factory::instance().create("&&&");
 	auto cmp = cif::compound_factory::instance().create("&&&");
 	REQUIRE(cmp == nullptr);
 	REQUIRE(cmp == nullptr);
 }
 }
+
+// --------------------------------------------------------------------
+// PDB2CIF tests
+
+TEST_CASE("pdb2cif_formula_weight")
+{
+	cif::compound_factory::instance().push_dictionary(gTestDir / "REA.cif");
+
+	cif::file a = cif::pdb::read(gTestDir / "pdb1cbs.ent.gz");
+	
+	auto fw = a.front()["entity"].find1<float>(cif::key("id") == 1, "formula_weight");
+	CHECK(std::abs(fw - 15581.802f) < 0.1f);
+
+	fw = a.front()["entity"].find1<float>(cif::key("id") == 2, "formula_weight");
+	CHECK(fw == 300.435f);
+
+	fw = a.front()["entity"].find1<float>(cif::key("id") == 3, "formula_weight");
+	CHECK(fw == 18.015f);
+}
\ No newline at end of file