Merge branch 'develop'

docs/exportdata.rst

@@ -0,0 +1,37 @@
+###############################
+Exporting data from ISA formats
+###############################
+
+We have provided a number of modules that allow you to export data from ISA formats to formats for well-known
+databases or services.
+
+----------------------------------
+Exporting ISA content to SampleTab
+----------------------------------
+To export metadata from SampleTab files (e.g. for EBI BioSamples database), you can do the following to export a
+ISA-Tab to SampleTab:
+
+.. code-block:: python
+
+    from isatools.convert import isatab2sampletab
+    with open('your/path/to/i_investigation.txt', 'r') as input_investigation_file:
+        with open('your/path/to/sampletab.txt', 'w') as output_sampletab_file:
+            isatab2sampletab.convert(input_investigation_file, output_sampletab_file)
+
+To export an ISA JSON file to SampleTab, you can do:
+
+.. code-block:: python
+
+    from isatools.convert import isatab2sampletab
+    with open('your/path/to/i_investigation.txt', 'r') as input_investigation_file:
+        with open('your/path/to/sampletab.txt', 'w') as output_sampletab_file:
+            isatab2sampletab.convert(input_investigation_file, output_sampletab_file)
+
+You can also dump SampleTab content directly from ISA Python objects:
+
+.. code-block:: python
+
+    from isatools import sampletab
+    with open('your/path/to/sampletab.txt', 'w') as output_sampletab:
+        # Note: ISA would be a previously loaded or constructed root Investigation object
+        sampletab.dump(ISA, output_sampletab)

docs/importdata.rst

@@ -99,4 +99,34 @@ tool from https://github.com/ISA-tools/mzml2isa and can be run as follows:
 .. code-block:: python
 
     from isatools.convert import mzml2isa
-    mzml2isa.convert('your/path/to/mzml/files/', 'tmp/', "My Study ID")
+    mzml2isa.convert('your/path/to/mzml/files/', 'tmp/', "My Study ID")
+
+
+--------------------------
+Importing SampleTab to ISA
+--------------------------
+To import metadata from SampleTab files (e.g. from EBI BioSamples database), you can do the following to import a
+SampleTab to ISA-Tab:
+
+.. code-block:: python
+
+    from isatools.convert import sampletab2isatab
+    with open('your/path/to/sampletab.txt', 'r') as input_sampletab:
+        sampletab2isatab.convert(input_sampletab, 'tmp/')
+
+To import a SampleTab to ISA JSON, you can do:
+
+.. code-block:: python
+
+    from isatools.convert import sampletab2json
+    with open('your/path/to/sampletab.txt', 'r') as input_sampletab:
+        with open('your/path/to/myjson.json', 'w') as output_json:'
+            sampletab2json.convert(input_sampletab, output_json)
+
+You can also load SampleTab content directly into ISA Python objects:
+
+.. code-block:: python
+
+    from isatools import sampletab
+    with open('your/path/to/sampletab.txt', 'r') as input_sampletab:
+        ISA = sampletab.load(input_sampletab)

docs/index.rst

@@ -19,6 +19,7 @@ flexible way to build and use ISA content, as well as provide utility functions
 #. :doc:`Downloading files stored in Github </github>`
 #. :doc:`Validating ISA-Tab and ISA JSON </validation>`
 #. :doc:`Importing data in ISA formats </importdata>`
+#. :doc:`Exporting data in ISA formats </exportdata>`
 #. :doc:`Known Issues </knownissues>`
 
 License
@@ -43,4 +44,5 @@ This code is licensed under the `CPAL License <https://raw.githubusercontent.com
    Downloading files stored in Github <github>
    Validating ISA-Tab and ISA JSON <validation>
    Importing data in ISA formats <importdata>
+   Exporting data from ISA formats <exportdata>
    Known Issues <knownissues>

docs/knownissues.rst

@@ -2,6 +2,11 @@
 Known issues
 ############
 
+isatools v0.7 package
+---------------------
+- Issues #101, #153 are still outstanding, as per below
+- SRA/ENA importer and Biocrates importer relies on XSLT2 processing only available with SAXON and requires .jar file to run
+
 isatools v0.6 package
 ---------------------
 - Issues #146, #101, #153 are still outstanding, as per below

isatools/__main__.py

@@ -18,17 +18,18 @@ def main(argv=None):
                                 )
 
     p.add_argument('-c', dest='cmd', help='isatools API command to run', required=True,
-                   choices=['isatab2json', 'json2isatab', 'isatab2sra', 'json2sra', 'zipisatab'])
-    p.add_argument('-i', dest='in_path', help='in folder (files will be read from here)', required=True)
-    p.add_argument('-o', dest='out_path', help='out folder (a new directory will be created here)', required=True)
-    p.add_argument('--version', action='version', version='isatools {}'.format("0.6"))
+                   choices=['isatab2json', 'json2isatab', 'sampletab2isatab', 'sampletab2json'])
+    p.add_argument('-i', dest='in_path', help='in  (files or directory will be read from here)', required=True)
+    p.add_argument('-o', dest='out_path', help='out (file will be written out here or written to directory if ISA-Tab '
+                                               'archive out)', required=True)
+    p.add_argument('--version', action='version', version='isatools {}'.format("0.7"))
     p.add_argument('-v', dest='verbose', help="show more output", action='store_true', default=False)
 
     args = p.parse_args(argv or sys.argv[1:])
 
     if args.verbose:
-        print("{} input path: {}".format(os.linesep, args.in_path))
-        print("output path: {}".format(args.out_path))
+        print("{} input: {}".format(os.linesep, args.in_path))
+        print("output: {}".format(args.out_path))
 
     if args.cmd == 'isatab2json':
         from isatools.convert import isatab2json
@@ -41,14 +42,16 @@ def main(argv=None):
         with open(args.in_path) as in_fp:
             json2isatab.convert(in_fp, args.out_path)
 
-    elif args.cmd == 'isatab2sra':
-        raise NotImplementedError  # TODO: Finish this
-
-    elif args.cmd == 'json2sra':
-        raise NotImplementedError  # TODO: Finish this
+    elif args.cmd == 'sampletab2isatab':
+        from isatools.convert import sampletab2isatab
+        with open(args.in_path) as in_fp:
+            sampletab2isatab.convert(in_fp, args.out_path)
 
-    elif args.cmd == 'zipisatab':
-        raise NotImplementedError  # TODO: Finish this
+    elif args.cmd == 'sampletab2json':
+        from isatools.convert import sampletab2json
+        with open(args.in_path) as in_fp:
+            with open(args.out_path) as out_fp:
+                sampletab2json.convert(in_fp, out_fp)
 
 if __name__ == '__main__':
     main()

isatools/config/json/default/schemas/data_schema.json

@@ -24,7 +24,8 @@
                 "Post Translational Modification Assignment File",
                 "Derived Array Data Matrix File",
                 "Free Induction Decay Data File",
-                "Metabolite Assignment File"
+                "Metabolite Assignment File",
+                "Array Data Matrix File"
             ]
         },
         "comments" : {

isatools/convert/isatab2sampletab.py

@@ -0,0 +1,14 @@
+from isatools import isatab, sampletab
+import logging
+
+logging.basicConfig(format='%(asctime)s %(levelname)s: %(message)s', level=logging.INFO)
+logger = logging.getLogger(__name__)
+
+
+def convert(source_inv_fp, target_fp):
+    """ Converter for ISA-Tab to SampleTab.
+    :param source_inv_fp: File descriptor of input investigation file
+    :param target_fp: File descriptor to write output SampleTab to (must be writeable)
+    """
+    ISA = isatab.load(source_inv_fp)
+    sampletab.dump(ISA, target_fp)

isatools/convert/isatab2sra.py

@@ -4,6 +4,7 @@
 import logging
 import json
 from io import StringIO
+from isatools.convert import isatab2json, json2sra
 
 
 def zipdir(path, zip_file):
@@ -20,11 +21,10 @@ def zipdir(path, zip_file):
 
 
 def convert(source_path, dest_path, sra_settings=None, validate_first=True):
-    from isatools.convert import isatab2json, json2sra
     isa_json = isatab2json.convert(source_path, validate_first=validate_first)
     isa_json_fp = StringIO(json.dumps(isa_json))
     isa_json_fp.name = "BII-S-3.json"
-    json2sra.convert2(isa_json_fp, dest_path, sra_settings=sra_settings, validate_first=False)
+    json2sra.convert(isa_json_fp, dest_path, sra_settings=sra_settings, validate_first=False)
     logging.info("Conversion complete...")
     buffer = BytesIO()
     if os.path.isdir(dest_path):

isatools/convert/json2sampletab.py

@@ -0,0 +1,14 @@
+from isatools import isajson, sampletab
+import logging
+
+logging.basicConfig(format='%(asctime)s %(levelname)s: %(message)s', level=logging.INFO)
+logger = logging.getLogger(__name__)
+
+
+def convert(source_json_fp, target_fp):
+    """ Converter for ISA-JSON to SampleTab.
+    :param source_json_fp: File descriptor of input ISA JSON file
+    :param target_fp: File descriptor to write output SampleTab to (must be writeable)
+    """
+    ISA = isajson.load(source_json_fp)
+    sampletab.dump(ISA, target_fp)

isatools/convert/json2sra.py

@@ -1,27 +1,11 @@
 from isatools import isajson, sra
-
-import glob
-import os
 import logging
 
 logging.basicConfig(format='%(asctime)s %(levelname)s: %(message)s', level=logging.INFO)
 logger = logging.getLogger(__name__)
 
 
-def convert(json_fp, path, config_dir=None):
-    """ Converter for ISA JSON to SRA.
-    :param json_fp: File pointer to ISA JSON input
-    :param path: Directory for output to be written
-    :param config_dir: path to JSON configuration
-    """
-    from isatools.convert import json2isatab, isatab2sra
-    json2isatab.convert(json_fp=json_fp, path=path, config_dir=config_dir)
-    isatab2sra.create_sra(path, path)
-    for f in glob.iglob(os.path.join(path, '*.txt')):  # remove generated isatab files
-        os.remove(f)
-
-
-def convert2(json_fp, path, config_dir=None, sra_settings=None, datafilehashes=None, validate_first=True):
+def convert(json_fp, path, config_dir=None, sra_settings=None, datafilehashes=None, validate_first=True):
     """ (New) Converter for ISA JSON to SRA.
     :param json_fp: File pointer to ISA JSON input
     :param path: Directory for output to be written
@@ -53,5 +37,5 @@ def convert2(json_fp, path, config_dir=None, sra_settings=None, datafilehashes=N
 
 If files in filehashes dict don't map 1:1 to files found in ISA JSON content, raise Exception
 
-json2sra.convert2(json_fp=open('/Users/dj/PycharmProjects/isa-api/copo.json'), path='/Users/dj/PycharmProjects/isa-api/tmp', sra_settings=)
+json2sra.convert(json_fp=open('/Users/dj/PycharmProjects/isa-api/copo.json'), path='/Users/dj/PycharmProjects/isa-api/tmp', sra_settings=)
 """

isatools/convert/mw2isa.py

@@ -1099,7 +1099,7 @@ def mw2isa_convert(**kwargs):
                 each_record.insert(1, taxonid)
 
             # Building the Investigation Object and its elements:
-            investigation = Investigation()
+            investigation = Investigation(identifier=studyid)
             investigation.comments.append(Comment(name="Primary Database", value="NIH Metabolomics Workbench"))
             investigation.comments.append(Comment(name="conversion date", value=str(date.today())))
             investigation.comments.append(Comment(name="conversion software", value="MW2ISA version 1.0"))
@@ -1923,6 +1923,7 @@ def mw2isa_convert(**kwargs):
             # ATTEMPTING TO WRITE INVESTIGATION FILE:
             try:
                 print("writing 'investigation information' to file...")
+                print(isatab.dumps(investigation))
                 isatab.dump(investigation, outputpath)
             except IOError:
                 print("Error: in main() method can\'t open file or write data")

isatools/convert/sampletab2isatab.py

@@ -0,0 +1,16 @@
+from isatools import sampletab, isatab
+import logging
+
+logging.basicConfig(format='%(asctime)s %(levelname)s: %(message)s', level=logging.INFO)
+logger = logging.getLogger(__name__)
+
+
+def convert(source_sampletab_fp, target_dir):
+    """ Converter for ISA-JSON to SampleTab.
+    :param source_sampletab_fp: File descriptor of input SampleTab file
+    :param target_dir: Path to write out ISA-Tab files to
+    """
+    ISA = sampletab.load(source_sampletab_fp)
+    isatab.dump(ISA, target_dir)
+
+

isatools/convert/sampletab2json.py

@@ -0,0 +1,16 @@
+from isatools import sampletab
+from isatools.isajson import ISAJSONEncoder
+import json
+import logging
+
+logging.basicConfig(format='%(asctime)s %(levelname)s: %(message)s', level=logging.INFO)
+logger = logging.getLogger(__name__)
+
+
+def convert(source_sampletab_fp, target_json_fp):
+    """ Converter for ISA-JSON to SampleTab.
+    :param source_sampletab_fp: File descriptor of input SampleTab file
+    :param target_json_fp: File descriptor to write output ISA JSON (must be writeable)
+    """
+    ISA = sampletab.load(source_sampletab_fp)
+    json.dump(ISA, fp=target_json_fp, cls=ISAJSONEncoder)

isatools/examples/validateISAtab.py

@@ -2,7 +2,7 @@
 
 # Inspired by validateSBML.py example from libSBML Python API
 
-from isatools.isatab import validate2
+from isatools.isatab import validate
 import sys
 import os
 
@@ -29,7 +29,7 @@ def main(args):
             numfiles += 1
         else:
             with open(args[i]) as fp:
-                report = validate2(fp)
+                report = validate(fp)
                 numerrors = len(report['errors'])
                 numwarnings = len(report['warnings'])
                 if numerrors > 0: