Merge pull request #1214 from wasade/demux_fetch_bug

Demux fetch bug

Author: adamrp

qiita_ware/demux.py

@@ -438,7 +438,7 @@ def to_ascii(demux, samples=None):
     for samp, idx, seq, qual, bc_ori, bc_cor, bc_err in fetch(demux, samples):
         seq_id = id_fmt % {'sample': samp, 'idx': idx, 'bc_ori': bc_ori,
                            'bc_cor': bc_cor, 'bc_diff': bc_err}
-        if qual is not None:
+        if qual != []:
             qual = qual.astype(np.uint8)
 
         yield formatter(seq_id, seq, qual)
@@ -516,7 +516,7 @@ def fetch(demux, samples=None, k=None):
         seqs = demux[pjoin(dset_paths['sequence'])][indices]
 
         # only yield qual if we have it
-        quals = repeat(None)
+        quals = repeat([])
         if demux.attrs['has-qual']:
             if len(indices) == 1:
                 if indices[0]:
@@ -531,8 +531,8 @@ def fetch(demux, samples=None, k=None):
         iter_ = zip(repeat(sample), np.arange(indices.size)[indices], seqs,
                     quals, bc_original, bc_corrected, bc_error)
 
-        for item in iter_:
-            yield item
+        for samp, idx, seq, qual, bc_ori, bc_cor, bc_err in iter_:
+            yield (samp, idx, seq, qual[:len(seq)], bc_ori, bc_cor, bc_err)
 
 
 def stats(demux):

qiita_ware/test/test_demux.py

@@ -331,6 +331,25 @@ def test_fetch(self):
         # implicitly tested with test_to_ascii
         pass
 
+    def test_fetch_qual_length_bug(self):
+        # fetch was not trimming qual to the length of the sequence resulting
+        # in qual scores for positions beyond the length of the sequence.
+        with tempfile.NamedTemporaryFile('r+', suffix='.fq',
+                                         delete=False) as f:
+            f.write(fqdata_variable_length)
+
+        self.to_remove.append(f.name)
+        to_hdf5(f.name, self.hdf5_file)
+
+        exp = [('a', [(b"@a_0 orig_bc=abc new_bc=abc bc_diffs=0\nxyz\n+\n"
+                       "ABC\n")]),
+               ('b', [(b"@b_0 orig_bc=abw new_bc=wbc bc_diffs=4\nqwe\n+\n"
+                       "DFG\n"),
+                      (b"@b_1 orig_bc=abw new_bc=wbc bc_diffs=4\nqwexx\n+\n"
+                       "DEF#G\n")])]
+
+        obs = [(s[0], list(s[1])) for s in to_per_sample_ascii(self.hdf5_file)]
+        self.assertEqual(obs, exp)
 
 seqdata = """>a_1 orig_bc=abc new_bc=abc bc_diffs=0
 x
@@ -370,5 +389,19 @@ def test_fetch(self):
 DEF
 """
 
+fqdata_variable_length = """@a_1 orig_bc=abc new_bc=abc bc_diffs=0
+xyz
++
+ABC
+@b_1 orig_bc=abw new_bc=wbc bc_diffs=4
+qwe
++
+DFG
+@b_2 orig_bc=abw new_bc=wbc bc_diffs=4
+qwexx
++
+DEF#G
+"""
+
 if __name__ == '__main__':
     main()