Make use of isotope label information in InChi descriptions when read…

…ing transition lists and .msp files

Sometimes isotope label information can be found in InChi descriptions, e.g. the "1D3" in "InChI=1S/C8H8O/c1-7(9)8-5-3-2-4-6-8/h2-6H,1H3/i1D3" which tells us that three of the hydrogens in C8H8O are deuterium. Previously Skyline did not look for these, which could lead to confusing mismatches between declared and calculated m/z values.

pwiz_tools/Skyline/Model/CustomMolecule.cs

@@ -102,6 +102,16 @@ public static Dictionary<string, string> FormatAccessionNumbers(string keysTSV,
             return keys;
         }
 
+        public static MoleculeAccessionNumbers Create(Dictionary<string, string> accessions)
+        {
+            if (accessions == null || accessions.Count == 0)
+            {
+                return EMPTY;
+            }
+
+            return new MoleculeAccessionNumbers(accessions);
+        }
+
         public MoleculeAccessionNumbers(string keysTSV, string inChiKey = null)
         {
             var keys = FormatAccessionNumbers(keysTSV, inChiKey);
@@ -115,6 +125,52 @@ public MoleculeAccessionNumbers(string keysTSV, string inChiKey = null)
             AccessionNumbers = ImmutableSortedList<string, string>.FromValues(keys, ACCESSION_TYPE_SORTER); 
         }
 
+        private static readonly Regex REGEX_INCHI_ISOTOPES = new Regex(@"\d+([A-Za-z]+\d+)", // Position, isotope, weight (or isotope and count, for D and T)
+            RegexOptions.CultureInvariant | RegexOptions.Compiled); // N.B. we ignore position, as we don't carry that much structure detail
+
+        // Look for labels buried in descriptions, e.g. InChi's /i section
+        public Dictionary<string, int> FindLabels()
+        {
+            var inchi = GetInChI();
+            if (string.IsNullOrEmpty(inchi))
+            {
+                return null; // No InChI at all
+            }
+
+            // e.g. InChI=1S/C8H8O/c1-7(9)8-5-3-2-4-6-8/h2-6H,1H3/i1D4 (replace 4 H with H')
+            // e.g. InChI=1S/C8H8O/c1-7(9)8-5-3-2-4-6-8/h2-6H,1H3/i1C13,2C13,3C13,4C13 (replace C with C' at positions 1,2,3, and 4)
+            var parts = inchi.Split('/');
+            if (parts.Length < 5)
+            {
+                return null; // No /i section
+            }
+
+            Dictionary<string, int> result = null;
+            var matches = REGEX_INCHI_ISOTOPES.Matches(parts[4]);
+
+            foreach (Match match in matches)
+            {
+                var isotope = match.Groups[1].Value;
+                var count = 1;
+                if (isotope.StartsWith(BioMassCalc.D) || isotope.StartsWith(BioMassCalc.T)) // e.g. "D4" in ".../i1D4" (replace 4 H with H') 
+                {
+                    if (!int.TryParse(isotope.Substring(1), out count)) // Get the count, if any e.g. 3 in "/i1T3"
+                    {
+                        count = 1;
+                    }
+                    isotope = isotope.Substring(0, 1);
+                }
+                if (Adduct.DICT_ADDUCT_ISOTOPE_NICKNAMES.TryGetValue(isotope, out var skylineIsotope)) // e.g. "C13" => "C'"
+                {
+                    result ??= new Dictionary<string, int>();
+                    result.TryGetValue(skylineIsotope, out var existing);
+                    result[skylineIsotope] = existing + count;
+                }
+            }
+
+            return result;
+        }
+
         public bool IsEmpty {
             get
             {

pwiz_tools/Skyline/Model/Lib/LibResources.resx

@@ -344,4 +344,7 @@
   <data name="NistLibraryBase_GetMod_Unknown_modification__0__at_line__1_" xml:space="preserve">
       <value>Unknown modification {0} at line {1}</value>
   </data>
+  <data name="NistLibraryBase_CreateCache_Missing_details_for__0__at_line__1___this_entry_will_be_ignored" xml:space="preserve">
+      <value>Missing details for {0} at line {1}, this entry will be ignored</value>
+  </data>
 </root>

pwiz_tools/Skyline/Model/Lib/Library.cs

@@ -2142,6 +2142,14 @@ public static SmallMoleculeLibraryAttributes Create(string moleculeName, string
             return Create(moleculeName, chemicalFormulaOrMassesString, inChiKey, otherKeys == null ? string.Empty : string.Join(@"\t", otherKeys.Select(kvp => kvp.Key + @":" + kvp.Value)));
         }
 
+        public static SmallMoleculeLibraryAttributes Create(string moleculeName, string chemicalFormulaOrMassesString,
+            IDictionary<string, string> accessions)
+        {
+            accessions.TryGetValue(MoleculeAccessionNumbers.TagInChiKey, out var inChiKey);
+            return Create(moleculeName, chemicalFormulaOrMassesString, inChiKey, 
+                string.Join(@"\t", accessions.Where(kvp => kvp.Key != MoleculeAccessionNumbers.TagInChiKey).Select(kvp => kvp.Key + @":" + kvp.Value)));
+        }
+
         public static SmallMoleculeLibraryAttributes Create(string moleculeName, string chemicalFormulaOrMassesString,
             string inChiKey, string otherKeys)
         {