Refactor portals (minor) and add more tests to improve coverage (#53)

Author: RJMW

README.rst

@@ -2,15 +2,15 @@ DIMSpy
 ======
 |Version| |Py versions| |Git| |Bioconda| |Build Status (Travis)| |Build Status (AppVeyor)| |License| |RTD doc| |codecov| |binder|
 
-Python package to process direct-infusion mass spectrometry-based metabolomics and lipidomics data
+Python package for processing direct-infusion mass spectrometry-based metabolomics and lipidomics data
 
-- **Documentation:** https://computational-metabolomics.github.io/dimspy
+- **Documentation:** https://dimspy.readthedocs.io/en/latest
 - **Source:** https://github.com/computational-metabolomics/dimspy
 - **Bug reports:** https://github.com/computational-metabolomics/dimspy/issues
 
 Installation
 ------------
-See the `Installation page <https://computational-metabolomics.github.io/dimspy/introduction.html#installation>`__ of
+See the `Installation page <https://dimspy.readthedocs.io/en/latest/introduction.html#installation>`__ of
 the `online documentation <https://computational-metabolomics.github.io/dimspy/>`__.
 
 
@@ -31,9 +31,10 @@ will help you to make the PR if you are new to `git`.
 
 Developers & Contributors
 -------------------------
- - Ralf J. M. Weber ([email protected]) - `University of Birmingham (UK) <http://www.birmingham.ac.uk/index.aspx>`_
+ - Ralf J. M. Weber ([email protected]) - `University of Birmingham (UK) <https://www.birmingham.ac.uk/staff/profiles/biosciences/weber-ralf.aspx>`_
  - Jiarui (Albert) Zhou ([email protected]) - `University of Birmingham (UK) <http://www.birmingham.ac.uk/index.aspx>`_, `HIT Shenzhen (China) <http://www.hitsz.edu.cn>`_
  - Thomas N. Lawson ([email protected]) - `University of Birmingham (UK) <http://www.birmingham.ac.uk/index.aspx>`_
+ - Martin R. Jones ([email protected]) - `Eawag  (Switzerland) <https://www.eawag.ch/en/aboutus/portrait/organisation/staff/profile/martin-jones/show/>`_
 
 
 License
@@ -63,7 +64,7 @@ Released under the GNU General Public License v3.0 (see `LICENSE file <https://g
    :target: https://www.gnu.org/licenses/gpl-3.0.html
 
 .. |RTD doc| image:: https://img.shields.io/badge/documentation-RTD-71B360.svg?style=flat&maxAge=3600
-   :target: https://computational-metabolomics.github.io/dimspy/
+   :target: https://dimspy.readthedocs.io/en/latest/
 
 .. |codecov| image:: https://codecov.io/gh/computational-metabolomics/dimspy/branch/master/graph/badge.svg
    :target: https://codecov.io/gh/computational-metabolomics/dimspy

dimspy/__main__.py

@@ -9,15 +9,15 @@
 from dimspy import __version__
 
 
-def map_delimiter(delimiter):
+def map_delimiter(delimiter):  # pragma: no cover
     seps = {"comma": ",", "tab": "\t"}
     if delimiter in seps:
         return seps[delimiter]
     else:
         return delimiter
 
 
-def main():
+def main():  # pragma: no cover
 
     print(("Executing dimspy version %s." % __version__))
 

dimspy/portals/mzml_portal.py

@@ -85,8 +85,7 @@ def peaklist(self, scan_id, function_noise="median"):
                 return pl
         return None
 
-    def peaklists(self, scan_ids, function_noise="median"):  # generator
-
+    def peaklists(self, scan_ids, function_noise="median"):
         if function_noise not in ["mean", "median", "mad"]:
             raise ValueError("select a function that is available [mean, median, mad]")
         run = pymzml.run.Reader(self.filename)
@@ -95,38 +94,33 @@ def peaklists(self, scan_ids, function_noise="median"):  # generator
         return pls
 
     def tics(self):
+        tic_values = collections.OrderedDict()
         run = pymzml.run.Reader(self.filename)
         for scan in run:
-            if scan["id"] == "TIC":
-                hit =  zip(*scan.peaks("raw"))[1]
-                run.info["file_object"].close()
-                return hit
+            tic_values[scan["id"]] = scan.TIC
         run.info["file_object"].close()
-        return
+        return tic_values
 
-    def injection_times(self):
-        injection_times = {}
+    def ion_injection_times(self):
+        iits = collections.OrderedDict()
         run = pymzml.run.Reader(self.filename)
         for scan in run:
-            injection_times[scan['id']] = None
-            for element in scan.xmlTree:
-                if "MS:1000927" == element.get('accession'):
-                    injection_times[scan['id']] = float(element.get("value"))
-                    break
-            if scan['id'] not in injection_times:
-                injection_times[scan['id']] = None
+            if "MS:1000927" in scan:
+                iits[scan['id']] = scan["MS:1000927"]
+            else:
+                iits[scan['id']] = None
         run.info["file_object"].close()
-        return injection_times
+        return iits
 
     def scan_dependents(self):
         l = []
         run = pymzml.run.Reader(self.filename)
         for scan in run:
             if type(scan["id"]) == int:
                 scan_id = scan["id"]
-                if "precursors" in list(scan.keys()):
+                if hasattr(scan, "precursors"):
                     spectrum_ref = None
-                    for element in scan.xmlTree:
+                    for element in scan.element:
                         for e in list(element.items()):
                             if e[0] == 'spectrumRef':
                                 spectrum_ref = int(e[1].split("scan=")[1])

dimspy/portals/thermo_raw_portal.py

@@ -123,20 +123,20 @@ def tics(self):
         tics = collections.OrderedDict()
         for scan_id in range(self.run.RunHeaderEx.FirstSpectrum, self.run.RunHeaderEx.LastSpectrum + 1):
             scan_stats = self.run.GetScanStatsForScanNumber(scan_id)
-            tics[scan_id].append(scan_stats.TIC)
+            tics[scan_id] = scan_stats.TIC
         return tics
 
-    def injection_times(self):
-        injection_times = collections.OrderedDict()
+    def ion_injection_times(self):
+        iits = collections.OrderedDict()
         for scan_id in range(self.run.RunHeaderEx.FirstSpectrum, self.run.RunHeaderEx.LastSpectrum + 1):
             extra_values = list(self.run.GetTrailerExtraInformation(scan_id).Values)
             extra_labels = list(self.run.GetTrailerExtraInformation(scan_id).Labels)
             for i, label in enumerate(extra_labels):
                 if "Ion Injection Time (ms):" == label:
-                    injection_times[scan_id] = float(extra_values[i])
-            if scan_id not in injection_times:
-                injection_times[scan_id] = None
-        return injection_times
+                    iits[scan_id] = float(extra_values[i])
+            if scan_id not in iits:
+                iits[scan_id] = None
+        return iits
 
     def scan_dependents(self):
         l = []

requirements.txt

@@ -1,7 +1,6 @@
 numpy==1.16.2
 scipy==1.1.0
 fastcluster==1.1.25
-plotly==2.5.1
 pymzml==2.4.2
 h5py==2.8.0
 pythonnet==2.4.0

tests/test_mzml_portal.py

@@ -0,0 +1,39 @@
+#!/usr/bin/env python
+#  -*- coding: utf-8 -*-
+
+
+import unittest
+import os
+from dimspy.portals.mzml_portal import Mzml
+
+
+def to_test_data(*args):
+    return os.path.join(os.path.dirname(os.path.realpath(__file__)), "data", "MTBLS79_subset", *args)
+
+def to_test_result(*args):
+    return os.path.join(os.path.dirname(os.path.realpath(__file__)), "test_results", *args)
+
+
+class MzmlPortalsTestCase(unittest.TestCase):
+
+    def test_mzml_portal(self):
+
+        run = Mzml(to_test_data("mzml", "batch04_QC17_rep01_262.mzML"))
+        self.assertEqual((run.run.get_spectrum_count(), run.run.get_spectrum_count()), (88, 88))
+        self.assertListEqual(list(run.headers().keys()), ['FTMS + p ESI w SIM ms [70.00-170.00]',
+                                                          'FTMS + p ESI w SIM ms [140.00-240.00]',
+                                                          'FTMS + p ESI w SIM ms [210.00-310.00]',
+                                                          'FTMS + p ESI w SIM ms [280.00-380.00]',
+                                                          'FTMS + p ESI w SIM ms [350.00-450.00]',
+                                                          'FTMS + p ESI w SIM ms [420.00-520.00]',
+                                                          'FTMS + p ESI w SIM ms [490.00-590.00]'])
+        self.assertListEqual(list(run.scan_ids().keys()), list(range(1,89)))
+        self.assertListEqual(list(run.tics().values())[0:2], [39800032.0, 38217892.0])
+        self.assertEqual(len(run.tics()), 88)
+        self.assertListEqual(list(run.ion_injection_times().values())[0:2], [40.433891296387, 40.094646453857])
+        self.assertEqual(len(run.ion_injection_times()), 88)
+        self.assertListEqual(run.scan_dependents(), [])
+        run.close()
+
+if __name__ == '__main__':
+    unittest.main()

tests/test_thermo_raw_portal.py

@@ -0,0 +1,38 @@
+#!/usr/bin/env python
+#  -*- coding: utf-8 -*-
+
+
+import unittest
+import os
+from dimspy.portals.thermo_raw_portal import ThermoRaw
+
+
+def to_test_data(*args):
+    return os.path.join(os.path.dirname(os.path.realpath(__file__)), "data", "MTBLS79_subset", *args)
+
+def to_test_result(*args):
+    return os.path.join(os.path.dirname(os.path.realpath(__file__)), "test_results", *args)
+
+
+class ThermoRawPortalsTestCase(unittest.TestCase):
+
+    def test_thermo_raw_portal(self):
+
+        run = ThermoRaw(to_test_data("raw", "batch04_QC17_rep01_262.RAW"))
+        self.assertListEqual(list(run.headers().keys()), ['FTMS + p ESI w SIM ms [70.00-170.00]',
+                                                          'FTMS + p ESI w SIM ms [140.00-240.00]',
+                                                          'FTMS + p ESI w SIM ms [210.00-310.00]',
+                                                          'FTMS + p ESI w SIM ms [280.00-380.00]',
+                                                          'FTMS + p ESI w SIM ms [350.00-450.00]',
+                                                          'FTMS + p ESI w SIM ms [420.00-520.00]',
+                                                          'FTMS + p ESI w SIM ms [490.00-590.00]'])
+        self.assertListEqual(list(run.scan_ids().keys()), list(range(1,89)))
+        self.assertListEqual(list(run.tics().values())[0:2], [39800032.0, 38217892.0])
+        self.assertEqual(len(run.tics()), 88)
+        self.assertListEqual(list(run.ion_injection_times().values())[0:2], [40.434, 40.095])
+        self.assertEqual(len(run.ion_injection_times()), 88)
+        self.assertListEqual(run.scan_dependents(), [])
+        run.close()
+
+if __name__ == '__main__':
+    unittest.main()

tests/test_tools.py

@@ -11,6 +11,7 @@
 
 
 import unittest
+import os
 import copy
 import numpy as np
 from dimspy.tools import *