translate: add check_start option and refactor, v.0.2.1

CHANGELOG

@@ -1,3 +1,5 @@
+v0.2.1:
+  * translate: add check_start option
 v0.2.0:
   * io: add option to write files automatically into an archive
   * io.gff: allow to store comment lines while reading

sugar/__init__.py

@@ -52,7 +52,7 @@
 The test suite can be run with ``sugar test``.
 """
 
-__version__ = '0.2.1-dev'
+__version__ = '0.2.1'
 
 from sugar.core.meta import Attr, Meta
 from sugar.core.fts import Feature, FeatureList, Location

sugar/core/translate.py

@@ -4,65 +4,69 @@
 import warnings
 
 
-def translate(seq, complete=False, gap='-', gap_after=2,
-              astop='X', warn=False, check_stop=False, tt=1):
+def translate(seq, *, complete=False, check_start=None, check_stop=False,
+              warn=False, astop='X', gap='-', gap_after=2, tt=1):
     """
     Translate a string or `.BioSeq` object into an amino acid string
 
-    :param bool complete: If set to ``True`` ignores stop codons
+    :param bool complete: If set to ``True`` ignores stop codons,
+        otherwise the translation is stopped before the first stop codon
+    :param bool check_start: Check that the first codon is a start codon,
+        default is False for ``complete=False`` otherwise True
+    :param bool check_stop: Check that the sequence ends with the first stop
+        codon, default is False
+    :param bool warn: Warn if the first codon might not be a start codon
+        (if ``check_start=True``) and warn for amigious stop codons,
+        default is False
+    :param str astop: Symbol for ambigious stop codons
     :param str gap: gap character, default ``'-'``, set to ``None``
        to raise an error for non nucleotide characters
     :param int gap_after: A single gap in the amino acis string is
         written after the first ``gap_after`` gaps in the
         nucleotide sequence and afterwards after each third gap,
-        defaults to 2
-    :param str astop: Symbol for ambigious stop codons
-    :param bool warn: Warn if start codon might not be a start codon and
-        warn for amigious stop codons for
-        ``complete=False``
-    :param bool check_stop: Check that last codon is a stop codon for
-        ``complete=False``
+        default is 2
+    :param int tt: the number of the translation table, default is 1
     """
     gc = gcode(tt)
     aas = []
-    codon = ''
     ngap = 0
-    check_start_codon = not complete
-    for nb in str(seq).replace('U', 'T'):
-        if nb == gap:
+    check_start = check_start if check_start is not None else not complete
+    codon = ''
+    for i, nt in enumerate(str(seq).replace('U', 'T')):
+        if nt == gap:
             ngap += 1
         else:
-            codon = codon + nb
+            codon = codon + nt
         if gap and gap_after is not None and ngap == gap_after:
             aas.append(gap)
             ngap -= 3
         if len(codon) == 3:
-            if check_start_codon:
+            if check_start:
+                check_start = False
                 if codon not in gc.starts and codon not in gc.astarts:
                     msg = (f'Codon {codon} is not a start codon {gc.starts} '
                            f'in genetic code #{gc.id}')
                     raise ValueError(msg)
-                if warn and codon not in gc.starts:
+                if codon not in gc.starts:
                     msg = f'Codon {codon} possibly is not a start codon.'
                     warnings.warn(msg)
-                check_start_codon = False
             try:
                 aa = gc.tt[codon]
             except (KeyError):
                 aa = 'X'
-            if not complete and codon in gc.stops:
-                break
             if codon in gc.astops:
                 aa = astop
-                if warn and not complete:
-                    msg = f'Codon {codon} might be a stop codon.'
-                    warnings.warn(msg)
+                if warn:
+                    warnings.warn(f'Codon {codon} might be a stop codon.')
+            if codon in gc.stops:
+                if check_stop and i < len(nt) - 1:
+                    msg = 'First stop codon is not at the end of the sequence.'
+                    raise ValueError(msg)
+                if not complete:
+                    break
             aas.append(aa)
             codon = ''
-    else:
-        if not complete and check_stop:
-            msg = (f'Last codon is not a stop codon {gc.stops} '
-                   f'in genetic code #{gc.id}')
-            raise ValueError(msg)
-
+    if check_stop and aa != 'X':
+        msg = f'Last codon is not a stop codon {gc.stops} in genetic code #{gc.id}'
+        raise ValueError(msg)
     return ''.join(aas)

sugar/tests/test_core_translate.py

@@ -9,7 +9,7 @@ def test_translate():
     s = 'CCC-AT-GAT-NCC--CCCCT---ANT-A--GGGN'
     aas = 'P-MX-PP-X-*G'
 
-    assert translate(s, complete=True, warn=True) == aas
+    assert translate(s, complete=True) == aas
     with pytest.warns(UserWarning, match='might be a stop'):
         assert translate(s[3:-3], warn=True) == aas[1:-2]
     assert translate(s[3:-3], warn=False) == aas[1:-2]