Merge pull request #52 from standage/fix/fasta

Fix missing ##FASTA directives, better handling of feature scores

tag/__init__.py

@@ -19,6 +19,7 @@
 from tag.range import Range
 from tag.reader import GFF3Reader
 from tag.writer import GFF3Writer
+from tag.score import Score
 from tag import cli
 from tag import select
 from tag import mrna

tag/feature.py

@@ -12,6 +12,7 @@
 from tag.directive import Directive
 from tag.range import Range
 from tag.sequence import Sequence
+from tag.score import Score
 
 
 class Feature(object):
@@ -59,10 +60,7 @@ def __init__(self, data):
         self._source = fields[1]
         self._type = fields[2]
         self._range = Range(int(fields[3]) - 1, int(fields[4]))
-        if fields[5] == '.':
-            self.score = None
-        else:
-            self.score = float(fields[5])
+        self._score = Score(fields[5])
         self._strand = fields[6]
         if fields[7] == '.':
             self.phase = None
@@ -78,15 +76,13 @@ def __init__(self, data):
 
     def __str__(self):
         """String representation of the feature, sans children."""
-        score = '.'
-        if self.score is not None:
-            score = "{:.3f}".format(self.score)
         phase = '.'
         if self.phase is not None:
             phase = str(self.phase)
         return '\t'.join([
             self.seqid, self.source, self.type, str(self.start + 1),
-            str(self.end), score, self.strand, phase, self.attributes
+            str(self.end), str(self._score), self.strand, phase,
+            self.attributes
         ])
 
     def __repr__(self):
@@ -236,6 +232,10 @@ def num_children(self):
     def fid(self):
         return self.get_attribute('ID')
 
+    @property
+    def score(self):
+        return self._score.value
+
     @property
     def slug(self):
         """

tag/score.py

@@ -0,0 +1,35 @@
+#!/usr/bin/env python
+#
+# -----------------------------------------------------------------------------
+# Copyright (C) 2017 Daniel Standage <daniel.standage@gmail.com>
+#
+# This file is part of tag (http://github.com/standage/tag) and is licensed
+# under the BSD 3-clause license: see LICENSE.
+# -----------------------------------------------------------------------------
+
+from __future__ import print_function
+import re
+
+
+class Score(object):
+
+    def __init__(self, datastr):
+        self._type = None
+        if datastr == '.':
+            self.value = None
+        elif re.search('^-*\d+$', datastr):
+            self.value = int(datastr)
+            self._type = int
+        else:
+            self.value = float(datastr)
+            self._type = float
+
+    def __str__(self):
+        if self.value is None:
+            return '.'
+        elif self._type == int:
+            return '{:d}'.format(self.value)
+        elif abs(self.value) < 1e6 and abs(self.value) > 1e-4:
+            return '{:1.3f}'.format(self.value)
+        else:
+            return '{:1.3E}'.format(self.value)

tag/writer.py

@@ -16,6 +16,7 @@
 import sys
 import tag
 from tag import Feature
+from tag import Sequence
 from tag import GFF3Reader
 
 
@@ -45,6 +46,7 @@ def __init__(self, instream, outfile='-'):
             self.outfile = tag.open(outfile, 'w')
         self.retainids = False
         self.feature_counts = defaultdict(int)
+        self._seq_written = False
 
     def __del__(self):
         if self.outfilename != '-' and not isinstance(self.outfile, StringIO):
@@ -62,4 +64,7 @@ def write(self):
                         feature.add_attribute('ID', fid)
                     else:
                         feature.drop_attribute('ID')
+            if isinstance(entry, Sequence) and not self._seq_written:
+                print('##FASTA', file=self.outfile)
+                self._seq_written = True
             print(repr(entry), file=self.outfile)

tests/test_score.py

@@ -0,0 +1,18 @@
+#!/usr/bin/env python
+#
+# -----------------------------------------------------------------------------
+# Copyright (C) 2017 Daniel Standage <daniel.standage@gmail.com>
+#
+# This file is part of tag (http://github.com/standage/tag) and is licensed
+# under the BSD 3-clause license: see LICENSE.
+# -----------------------------------------------------------------------------
+
+from tag import Score
+
+
+def test_basic():
+    for score in ['.', '10', '12.345', '-11', '-98765.432', '-4.555E+09']:
+        assert str(Score(score)) == score
+    assert str(Score('10.0')) == '10.000'
+    assert str(Score('1.32e12')) == '1.320E+12'
+    assert str(Score('1.2e-16')) == '1.200E-16'

tests/test_writer.py

@@ -46,3 +46,12 @@ def test_write_file():
     with open(output, 'r') as testout:
         testoutput2 = testout.read()
     assert testoutput1 == testoutput2, (testoutput1, testoutput2)
+
+
+def test_write_minimus():
+    reader = GFF3Reader(tag.pkgdata('minimus.gff3'))
+    output = StringIO()
+    writer = GFF3Writer(reader, output)
+    writer.write()
+
+    assert output.getvalue() == tag.pkgdata('minimus.gff3').read()

tests/testdata/minimus.gff3

@@ -0,0 +1,6 @@
+##gff-version 3
+##sequence-region minimus 1 10
+minimus	vim	region	4	6	9000	+	.	Note=lolwut
+##FASTA
+>minimus
+GATTACANNN