Merge branch 'master' of https://github.com/vdbwrair/ngs_mapper

README.rst

@@ -1,6 +1,10 @@
+.. image:: https://zenodo.org/badge/doi/10.5281/zenodo.46716.svg
+   :target: http://dx.doi.org/10.5281/zenodo.46716
+
 .. image:: https://badge.waffle.io/VDBWRAIR/ngs_mapper.png?label=ready&title=Ready 
- :target: https://waffle.io/VDBWRAIR/ngs_mapper
- :alt: 'Stories in Ready'
+    :target: https://waffle.io/VDBWRAIR/ngs_mapper
+    :alt: 'Stories in Ready'
+
 .. image:: https://readthedocs.org/projects/ngs_mapper/badge/?version=latest
     :target: http://ngs_mapper.readthedocs.org/en/latest/
     :alt: Documentation Status

ngs_mapper/data.py

@@ -126,7 +126,7 @@ def platform_for_read( filepath ):
     fh, ext = file_handle(filepath)
     # fix sanger abi extension issue
     if ext == 'ab1':
-        ext = 'abi'
+        return 'Sanger'
     try:
         # First record in readfile
         try:

ngs_mapper/file_formats.py

@@ -4,6 +4,9 @@
 from glob import glob
 import gzip
 from Bio import SeqIO
+from . import log
+
+logger = log.setup_logger(__name__, log.get_config())
 
 {
     'sff' : 'sff',
@@ -18,25 +21,28 @@ def convert_sff(dir):
     sff_paths = find_ext('sff')(dir)
     outnames = map(swap_ext('fastq'), sff_paths)
     def wrapped_conv(a, b):
-        print "Converting %s to fastq into %s" % (a, b)
+        logger.info('Converting {0} to {1}'.format(a, b))
         n = 0
         try:
             n=sffs_to_fastq([a], b, trim=True)
         except AssertionError as e:
             pass
-        print "%s reads converted" % n
+        logger.info("{0} reads converted".format(n))
         return n
     return sum(map(wrapped_conv, sff_paths, outnames))
 
 def convert_ab1(dir):
-    for abi in find_ext('gz')(dir):
-        SeqIO.convert(abi, 'abi', swap_ext('fastq')(abi), 'fastq')
+    for abi in find_ext('ab1')(dir):
+        dest = swap_ext('fastq')(abi)
+        logger.info('Converting {0} to {1}'.format(abi, dest))
+        SeqIO.convert(abi, 'abi', dest, 'fastq')
 
 def convert_gzips(dir):
     for gz in find_ext('gz')(dir):
         dest = drop_ext(gz)
         with gzip.open( gz, 'rb' ) as input:
             with open(dest, 'w') as output:
+                logger.info('Unpacking {0} to {1}'.format(gz, dest))
                 output.write(input.read())
 
 def convert_formats(dir):

ngs_mapper/tests/test_data.py

@@ -311,10 +311,8 @@ def test_identifies_sanger_reads(self, mseqio):
             mock.Mock(id='aZ_09-', seq='ATGC')
         ]
         for record in records:
-            mseqio.parse.return_value = iter([record])
             r = data.platform_for_read('/path/to/read.ab1')
             self.assertEqual('Sanger', r)
-            self.assertEqual('abi', mseqio.parse.call_args[0][1])
 
     def test_identifies_miseq_reads(self, mseqio):
         #@<instrument>:<run number>:<flowcell ID>:<lane>:<tile>:<x-pos>:<y-pos> <read>:<is filtered>:<control number>:<index sequence>