Add support for reading chemical formulas with numeric subscripts, e.…

…g. C₁₀H₁₄N₅O₈P reported anonymously via exception web
ProteoWizard · Nov 14, 2023 · 99e8470 · 99e8470
1 parent 6096a59
commit 99e8470
Show file tree

Hide file tree

Showing 3 changed files with 19 additions and 2 deletions.
diff --git a/pwiz_tools/Shared/CommonUtil/Chemistry/Formula.cs b/pwiz_tools/Shared/CommonUtil/Chemistry/Formula.cs
@@ -309,7 +309,17 @@ void CloseOutCurrentElement()
                 // Allow apostrophe for heavy isotopes (e.g. C' for 13C)
                 else if (!Char.IsWhiteSpace(ch))
                 {
-                    currentElement = currentElement + ch;
+                    // Watch out for unicode subscripts
+                    var n = ch - '\u2080'; // Subscript zero '₀'
+                    if (n >= 0 && n <= 9)
+                    {
+                        currentQuantity = (currentQuantity ?? 0) * 10 + n;
+                    }
+                    else
+                    {
+                        // Presumably part of an element description
+                        currentElement = currentElement + ch;
+                    }
                 }
             }
             CloseOutCurrentElement(); // Finish up the last element

diff --git a/pwiz_tools/Skyline/Test/MassCalcTest.cs b/pwiz_tools/Skyline/Test/MassCalcTest.cs
@@ -311,9 +311,10 @@ public void TestGetIonFormula()
             sequenceMassCalc.AddStaticModifications(new[] { labelLaK });
             Assert.AreEqual(294.033, sequenceMassCalc.GetPrecursorMass("K"), .1);
             Assert.AreEqual("C'6H14LaN'2O2", sequenceMassCalc.GetMolecularFormula("K").ToString());
-            
+
             // Check our ability to handle strangely constructed chemical formulas, and preserve nonstandard order
             Assert.AreEqual("C12H9S2", ParsedMolecule.Create("C12H9S2P0").ToString()); // P0 is weird, drop it
+            Assert.AreEqual("C12H9S2", ParsedMolecule.Create("C\u2081\u2082H\u2089S\u2082P\u2080").ToString()); // Same thing, unicode subscripts
             Assert.AreEqual("C12H9S2P1", ParsedMolecule.Create("C12H9S2P1").ToString()); // P1 is weird, but preserve it
             Assert.AreEqual("H9C12P", ParsedMolecule.Create("H9C12S0P").ToString()); // S0 is weird, and not at end
         }

diff --git a/pwiz_tools/Skyline/TestFunctional/PasteMoleculesTest.cs b/pwiz_tools/Skyline/TestFunctional/PasteMoleculesTest.cs
@@ -123,6 +123,7 @@ private static string ToLocalText(string text)
         const string caffeineSMILES = "Cn1cnc2n(C)c(=O)n(C)c(=O)c12";
         const string caffeineKEGG = "C07481";
         const string caffeineFormula = "C8H10N4O2";
+        const string caffeineFormulaUnicode = "C\u2088H\u2081\u2080N\u2084O\u2082"; // Unicode subscripts
         const string caffeineFragment = "C6H5N2O"; // Not really a known fragment of caffeine
 
         const double precursorMzAtZNeg2 = 96.0329118;
@@ -645,6 +646,11 @@ private void TestLegitimatePaste(string line2start, string[] fullColumnOrder)
             {
                 docEmpty = NewDocument();
                 line1 = BuildTestLine(imTypeIsDrift);
+                if (imTypeIsDrift)
+                {
+                    // Nothing to do with imType, just want to alternate styles here
+                    line1 = line1.Replace(caffeineFormula+"\t", caffeineFormulaUnicode + "\t"); // Test with unicode subscript numbers
+                }
                 var expectedIM = imTypeIsDrift ? precursorDT : compensationVoltage;
                 double? expectedCV = imTypeIsDrift ? (double?)null : compensationVoltage;
                 var expectedTypeIM = imTypeIsDrift ? eIonMobilityUnits.drift_time_msec : eIonMobilityUnits.compensation_V;