Merge pull request #32 from ECRL/dev

Addition to conversion tools, update to database creation function

docs/usage/tools.md

@@ -6,9 +6,7 @@ ECNet databases are comma-separated value (CSV) formatted files that provide inf
 
 Our [databases](https://github.com/ECRL/ECNet/tree/master/databases) directory on GitHub contains databases for cetane number, cloud point, kinetic viscosity, pour point and yield sooting index, as well as a database template.
 
-You can create an ECNet-formatted database with molecule names or SMILES and (optionally) target values. The following programs must be installed for you to do so:
-- [Open Babel](http://openbabel.org/wiki/Main_Page) software
-- [Java JRE](https://www.oracle.com/technetwork/java/javase/downloads/jre8-downloads-2133155.html) version 6 and above
+You can create an ECNet-formatted database with molecule names or SMILES and (optionally) target values. [Java JRE](https://www.oracle.com/technetwork/java/javase/downloads/jre8-downloads-2133155.html) version 6 and above is required to create a database.
 
 Supplied names or SMILES must exist in a text file, one entry per line:
 ```
@@ -70,6 +68,8 @@ smiles = get_smiles('Molecule Name')
 
 ### SMILES string to MDL Molfile
 
+Creating an MDL file requires [Open Babel](http://openbabel.org/wiki/Main_Page) to be installed.
+
 A text file containing SMILES strings, one per line, is required:
 
 ```
@@ -100,6 +100,19 @@ from ecnet.tools.conversions import mdl_to_descriptors
 mdl_to_descriptors('molfile.md', 'descriptors.csv')
 ```
 
+### SMILES string to QSPR descriptors
+
+A text file containing SMILES strings, one per line, is required, as well as an installation of [Java JRE](https://www.oracle.com/technetwork/java/javase/downloads/jre8-downloads-2133155.html).
+
+*Note: the file extension for the text file containing SMILES strings must be ".smi"
+
+```python
+from ecnet.tools.conversions import smiles_to_descriptors
+
+# A CSV file with descriptors is generated
+smiles_to_descriptors('molecules.smi', 'descriptors.csv')
+```
+
 ## ECNet .prj file usage
 
 Once an ECNet project has been created, the resulting .prj file can be used to predict properties for new molecules. A text file containing names or SMILES strings of new molecules, one per line, is required in addition to the .prj file.

ecnet/tools/conversions.py

@@ -40,6 +40,77 @@ def get_smiles(name):
     return [m.isomeric_smiles for m in get_compounds(name, 'name')]
 
 
+def smiles_to_descriptors(smiles_file, descriptors_csv, fingerprints=False):
+    '''Generates QSPR descriptors from supplied SMILES file using
+    PaDEL-Descriptor
+
+    Args:
+        smiles_file (str): path to source SMI file
+        descriptors_csv (str): path to resulting CSV file w/ descriptors
+        fingerprints (bool): if True, generates molecular fingerprints instead
+            of QSPR descriptors
+
+    Returns:
+        list: list of dicts, where each dict is a molecule populated with
+            descriptor names (keys) and values
+    '''
+
+    if which('java') is None:
+        raise ReferenceError(
+            'Java JRE 6+ not found (required for PaDEL-Descriptor)'
+        )
+
+    is_smi = compile(r'.*\.smi$', IGNORECASE)
+    if is_smi.match(smiles_file) is None:
+        raise ValueError('SMILES file must have a SMI extension: {}'.format(
+            smiles_file
+        ))
+
+    dn = open(devnull, 'w')
+    for attempt in range(3):
+        try:
+            if fingerprints:
+                call([
+                    'java',
+                    '-jar',
+                    _PADEL_PATH,
+                    '-fingerprints',
+                    '-retainorder',
+                    '-convert3d',
+                    '-retain3d',
+                    '-dir',
+                    smiles_file,
+                    '-file',
+                    descriptors_csv
+                ], stdout=dn, stderr=dn, timeout=15)
+                break
+            else:
+                call([
+                    'java',
+                    '-jar',
+                    _PADEL_PATH,
+                    '-2d',
+                    '-3d',
+                    '-retainorder',
+                    '-convert3d',
+                    '-retain3d',
+                    '-dir',
+                    smiles_file,
+                    '-file',
+                    descriptors_csv
+                ], stdout=dn, stderr=dn, timeout=15)
+                break
+        except Exception as e:
+            if attempt == 2:
+                raise e
+            else:
+                continue
+
+    with open(descriptors_csv, 'r', encoding='utf-8') as desc_file:
+        reader = DictReader(desc_file)
+        return [row for row in reader]
+
+
 def smiles_to_mdl(smiles_file, mdl_file):
     '''Invoke Open Babel to generate an MDL file containing all supplied
     molecules; requires Open Babel to be installed externally
@@ -71,7 +142,7 @@ def smiles_to_mdl(smiles_file, mdl_file):
                 '-O',
                 mdl_file,
                 '--gen3D'
-            ], stdout=dn, stderr=dn, timeout=3600)
+            ], stdout=dn, stderr=dn, timeout=15)
             break
         except Exception as e:
             if attempt == 2:
@@ -115,7 +186,7 @@ def mdl_to_descriptors(mdl_file, descriptors_csv, fingerprints=False):
                     mdl_file,
                     '-file',
                     descriptors_csv
-                ], stdout=dn, stderr=dn, timeout=600)
+                ], stdout=dn, stderr=dn, timeout=15)
                 break
             else:
                 call([
@@ -130,7 +201,7 @@ def mdl_to_descriptors(mdl_file, descriptors_csv, fingerprints=False):
                     mdl_file,
                     '-file',
                     descriptors_csv
-                ], stdout=dn, stderr=dn, timeout=600)
+                ], stdout=dn, stderr=dn, timeout=15)
                 break
         except Exception as e:
             if attempt == 2:

ecnet/tools/database.py

@@ -14,14 +14,12 @@
 from os import remove
 
 # ECNet imports
-from ecnet.tools.conversions import get_smiles, smiles_to_mdl,\
-    mdl_to_descriptors
+from ecnet.tools.conversions import get_smiles, smiles_to_descriptors
 
 
 def create_db(input_txt, output_name, id_prefix='', targets=None, form='name',
-              smiles_file='mols.smi', mdl_file='mols.mdl',
-              desc_file='descriptors.csv', clean_up=True, fingerprints=False,
-              extra_strings={}):
+              smiles_file='mols.smi', desc_file='descriptors.csv',
+              clean_up=True, fingerprints=False, extra_strings={}):
     '''Create an ECNet-formatted database from either molecule names or SMILES
     strings
 
@@ -32,9 +30,8 @@ def create_db(input_txt, output_name, id_prefix='', targets=None, form='name',
         id_prefix (str): prefix for DATAID column entries, if supplied
         targets (str): path to file containing target values (optional)
         form (str): `name` or `smiles` for selecting input format
-        smiles_file (str): if input format is not SMILES, this is the name of
-            a temporary .smi file containing SMILES strings
-        mdl_file (str): name of MDL file generated by Open Babel
+        smiles_file (str): name of temporary .smi file containing SMILES
+            strings
         desc_file (str): name of descriptors file generated by PaDEL-Descriptor
         clean_up (bool): if True, cleans up all files generated during
             processing except for the input text files and output database
@@ -66,7 +63,9 @@ def create_db(input_txt, output_name, id_prefix='', targets=None, form='name',
                                      len(input_names), len(input_data)
                                  ))
             del extra_strings['Compound Name']
-        smiles_file = input_txt
+        with open(smiles_file, 'w') as smi_file:
+            for d in input_data:
+                smi_file.write(d + '\n')
 
     else:
         raise ValueError('Unknown `format` argument: {}'.format(form))
@@ -83,8 +82,7 @@ def create_db(input_txt, output_name, id_prefix='', targets=None, form='name',
     else:
         target_data = [0 for _ in range(len(input_data))]
 
-    smiles_to_mdl(smiles_file, mdl_file)
-    desc = mdl_to_descriptors(mdl_file, desc_file, fingerprints)
+    desc = smiles_to_descriptors(smiles_file, desc_file, fingerprints)
     desc_keys = list(desc[0].keys())
     try:
         desc_keys.remove('Name')
@@ -135,9 +133,7 @@ def create_db(input_txt, output_name, id_prefix='', targets=None, form='name',
             wr.writerow(row)
 
     if clean_up:
-        if form != 'smiles':
-            remove(smiles_file)
-        remove(mdl_file)
+        remove(smiles_file)
         remove(desc_file)