fix: reject gene ids that split in grRules in getModelFromHomology

src/reconstruction/homology/getModelFromHomology.m

@@ -93,6 +93,11 @@
 % unneccesary old genes, the models should not have "or"-relations in their
 % grRules (use expandModel).
 %
+% The gene ids of the new organism in blastStructure become its gene ids in
+% the draft model, so they cannot contain whitespace, parentheses, "|" or
+% "&". FASTA headers from UniProt, such as "sp|P33284|HXK_KLULA", should be
+% reduced to the accession first.
+%
 % The resulting draft model contains only reactions associated with
 % orthologous genes, plus the gene-free ones with keepGeneFree. With
 % complexPolicy "flag", the old (original) genes involved in "and" relations
@@ -423,6 +428,17 @@
     finalMappings{i}=finalMappings{i}(usedNewGenes,:);
 end
 
+%Gene ids end up in grRules, which are split at whitespace and parentheses,
+%and which COBRA also splits at "|" and "&"
+unusable=allGenes{1}(~cellfun(@isempty,regexp(allGenes{1},'[\s()|&]','once')));
+if ~isempty(unusable)
+    error('RAVEN:badInput',['%d gene ids of "%s" cannot be written into a grRule, '...
+        'because they contain whitespace, parentheses, "|" or "&": %s. Map them '...
+        'to gene ids first, or use the accession alone for UniProt headers such '...
+        'as "sp|P33284|HXK_KLULA".'],numel(unusable),getModelFor,...
+        strjoin(unusable(1:min(3,end))',', '));
+end
+
 %Now is it time to choose which reactions should be included from which
 %models. If there is a preferred order specified then each gene can only
 %result in reactions from one model, otherwise they should all be included

tests/function_tests/tReconstruction.m

@@ -85,6 +85,13 @@ function getModelFromHomologyComplexPolicyAndOptions(testCase)
             testCase.verifyEqual(flag.rxnNotes{1}, 'Included by getModelFromHomology');
         end
 
+        function getModelFromHomologyRejectsGeneIdsThatSplit(testCase)
+            src = testCase.model; src.id = 'srcModel';
+            newGenes = cellstr(strcat('sp|P', string(1:10)', '|X_BUG'));
+            bs = makeFakeBlastStructure([src.genes(1:10), newGenes], 'srcModel', 'tgtOrg');
+            testCase.verifyError(@() getModelFromHomology({src}, bs, 'tgtOrg'), 'RAVEN:badInput');
+        end
+
         function getModelFromHomologyRejectsUnknownComplexPolicy(testCase)
             src = testCase.model; src.id = 'srcModel';
             bs = makeFakeBlastStructure([src.genes(1:10), strcat('t_', src.genes(1:10))], 'srcModel', 'tgtOrg');