Allow celltypist training to run when the job has more than one input, provided they are eventually merged

Author: bbimber

singlecell/resources/chunks/TrainCelltypist.R

@@ -1,6 +1,9 @@
-for (datasetId in names(seuratObjects)) {
-    printName(datasetId)
-    seuratObj <- readRDS(seuratObjects[[datasetId]])
+if (length(names(seuratObjects)) > 1) {
+    stop('Training expects a single input. Either include a merge step upstream or run jobs on individual input files')
+}
 
-    RIRA::TrainCellTypist(seuratObj, labelField = labelField, minCellsPerClass = minCellsPerClass, modelFile = modelFile, tempFileLocation = '/work')
-}
+datasetId <- names(seuratObjects)[[1]]
+printName(datasetId)
+seuratObj <- readRDS(seuratObjects[[datasetId]])
+
+RIRA::TrainCellTypist(seuratObj, labelField = labelField, minCellsPerClass = minCellsPerClass, modelFile = modelFile, tempFileLocation = '/work')

singlecell/src/org/labkey/singlecell/pipeline/singlecell/TrainCelltypist.java

@@ -65,14 +65,15 @@ public boolean createsSeuratObjects()
         return false;
     }
 
-    @Override
-    public void init(SequenceOutputHandler.JobContext ctx, List<SequenceOutputFile> inputFiles) throws PipelineJobException
-    {
-        if (inputFiles.size() > 1)
-        {
-            throw new PipelineJobException("Celltypist train step expects this job to have a single input. Consider selecting the option to run jobs individually instead of merged");
-        }
-    }
+    //@Override
+    //public void init(SequenceOutputHandler.JobContext ctx, List<SequenceOutputFile> inputFiles) throws PipelineJobException
+    //{
+    //    //NOTE: a valid use-case is to merge/filter many inputs, produce one object, and train, so dont perform this check.
+    //    if (inputFiles.size() > 1)
+    //    {
+    //        throw new PipelineJobException("Celltypist train step expects this job to have a single input. Consider selecting the option to run jobs individually instead of merged");
+    //    }
+    //}
 
     @Override
     public Output execute(SequenceOutputHandler.JobContext ctx, List<SeuratObjectWrapper> inputObjects, String outputPrefix) throws PipelineJobException