Fix output directory for genomic references and better precondition c…

…hecks

rnaseq_pipeline/targets.py

@@ -10,13 +10,17 @@ class RsemReference(luigi.Target):
     """
     Represents the target of rsem-prepare-reference script.
     """
-    def __init__(self, prefix, taxon):
-        self.prefix = prefix
+    def __init__(self, path, taxon):
+        self.path = path
         self.taxon = taxon
 
+    @property
+    def prefix(self):
+        return join(self.path, '{}_0'.format(self.taxon))
+
     def exists(self):
-        exts = ['grp', 'ti', 'seq', 'chrlist']
-        return all(exists(join(self.prefix, '{}_0.{}'.format(self.taxon, ext)))
+        exts = ['chrlist', 'grp', 'idx.fa', 'ng2.idx.fa', 'seq', 'ti', 'transcripts.fa']
+        return all(exists(self.prefix + '.' + ext)
                 for ext in exts)
 
 class GemmaDatasetPlatform(luigi.Target):

rnaseq_pipeline/tasks.py

@@ -194,9 +194,11 @@ class PrepareReference(ScheduledExternalProgramTask):
     def input(self):
         genome_dir = join(cfg.OUTPUT_DIR, cfg.GENOMES, self.reference_id)
         gtf_files = glob(join(genome_dir, '*.gtf'))
-        fasta_files = glob(join(genome_dir, '*.fn?a'))
+        fasta_files = glob(join(genome_dir, '*.f*a')) # FIXME: this pattern is too broad
         if len(gtf_files) != 1:
-            raise ValueError('Only one GTF file is expected in {}.'.format(genome_dir))
+            raise ValueError('Exactly one GTF file is expected in {}.'.format(genome_dir))
+        if len(fasta_files) < 1:
+            raise ValueError('At least one FASTA (with .fa or .fna extension) file is expected in {}.'.format(genome_dir))
         return [luigi.LocalTarget(gtf_files[0]),
                 [luigi.LocalTarget(f) for f in fasta_files]]
 
@@ -212,12 +214,12 @@ def program_args(self):
 
         args.extend([t.path for t in genome_fasta])
 
-        args.append(join(self.output().prefix))
+        args.append(self.output().prefix)
 
         return args
 
     def run(self):
-        os.makedirs(self.output().prefix, exist_ok=True)
+        os.makedirs(self.output().path, exist_ok=True)
         return super().run()
 
     def output(self):