Added missing docstrings.

OrthoEvol/Manager/database_management.py

@@ -92,11 +92,13 @@ def download_blast_database(self, database_name="refseq_rna", v5=True, set_blast
 
         For other types of blast data, please see the NCBIREADME.md file.
 
-        :param database_name:  A string that represents a pattern in the files of interest.
-        :type database_name:  string.
+        :param database_name:  A string that represents a pattern in the files of interest, defaults to "refseq_rna"
+        :type database_name:  str, optional
+        :param v5: A flag that determines which version of blastdb to use, defaults to True
+        :type v5: bool, optional
         :param set_blastdb:  A flag that determines whether the BLASTDB environment
                             variable is automatically set.
-        :type set_blastdb:  bool.
+        :type set_blastdb:  bool, optional
         """
         # <path>/<user or basic_project>/databases/NCBI/blast/db/<database_name>
         dl_path = Path(self.database_path) / Path("NCBI") / Path("blast") / Path("db")
@@ -333,8 +335,7 @@ def PROJECTS(self, **kwargs):
         print(self)
         return {}, {}
 
-    def full(self, NCBI, ITIS, Projects=None,
-             configure_flag=None, archive_flag=None, delete_flag=None, project_flag=None, _path=None):
+    def full(self, NCBI, ITIS, Projects=None, configure_flag=None, archive_flag=None, delete_flag=None, project_flag=None, _path=None):
         """
         The most generalized strategy available.  This configures everything.  The 3 primary flags (configure, archive,
         and delete) will be passed down to the more specific strategies, which will inherit these values unless

OrthoEvol/Orthologs/Blast/blast.py

@@ -158,7 +158,23 @@ def select_method(method=1):
         return blastn_parameters, query_config
 
     def blastn_wrapper(self, gene, organism, parameters, xml_path, gene_path):
-        """Run the Ncbiblastncommandline wrapper modified by this package."""
+        """Use NCBI's blastn wrapper to run blast.
+
+        The function includes a try/except to ensure that errors are caught
+        and that if a blast stops while blasting, incomplete files will be
+        removed.
+
+        :param gene: The input gene for the blast run.
+        :type gene: str
+        :param organism: The organism to retrieve a hit for.
+        :type organism: str
+        :param parameters: A dictionary of blastn parameters.
+        :type parameters: dict
+        :param xml_path: The path to the xml output file.
+        :type xml_path: str
+        :param gene_path: The path to the gene's directory.
+        :type gene_path: str
+        """
         try:
             self.blastn_log.info('Blast run has started.')
             start_time = self.get_time()
@@ -207,7 +223,8 @@ def blastn_wrapper(self, gene, organism, parameters, xml_path, gene_path):
             raise
 
     def configure(self, query_accessions, query_organism, auto_start=False):
-        """This method configures everything for our BLAST workflow.
+        """Configure the BLAST workflow.
+
         It configures the accession file, which works with interrupted Blasts.
         It configures a gene_list for blasting the right genes.
 
@@ -402,7 +419,7 @@ def runblast(self, genes=None, query_organism=None, pre_configured=False):
                             self.blastn_log.error('%s was deleted' % xml)
 
                     else:
-                         # Set up blast parameters
+                        # Set up blast parameters
                         query_seq_path = str(gene_path / Path('temp.fasta'))
                         # Add blastn parameters for each method to dict
                         if self.method == 1:

OrthoEvol/Orthologs/Blast/blastn_wrapper.py

@@ -99,7 +99,8 @@ def __init__(self, cmd="blastn", **kwargs):
                     """Discontiguous MegaBLAST template type (string).
                     Allowed values: 'coding', 'coding_and_optimal' or 'optimal'
                     Requires: template_length.""",
-                    checker_function=lambda value: value in ['coding', 'coding_and_optimal', 'optimal'],
+                    checker_function=lambda value: value in [
+                        'coding', 'coding_and_optimal', 'optimal'],
                     equate=False),
             _Option(["-template_length", "template_length"],
                     """Discontiguous MegaBLAST template length (integer).
@@ -128,4 +129,4 @@ def __init__(self, cmd="blastn", **kwargs):
                     """,
                     equate=False),
         ]
-        _NcbiblastMain2SeqCommandline.__init__(self, cmd, **kwargs)
+        _NcbiblastMain2SeqCommandline.__init__(self, cmd, **kwargs)

OrthoEvol/Tools/ftp/baseftp.py

@@ -13,16 +13,17 @@ class BaseFTPClient(object):
     """
 
     def __init__(self, ftpsite, user, password, keepalive=False, debug_lvl=0):
-        """Connect to a ftp site using an email address.
-
-        :param ftpsite: Address of the ftp site you want to connect to.
-        :param email: Input your email address.
-        :param keepalive: Flag to determine whether to keepalive the connection
-                          (Default value = False)
-        :type keepalive: bool
-        :param debug_lvl: Verbosity level for debugging ftp connection
-                          (Default value = 0)
-        :type debug_lvl: int
+        """Connect to a ftp site using a username and password.
+
+        :param ftpsite: The url or http address of the ftp site you want to connect to.
+        :param user: The name of the user that will log in.
+        :type user: str
+        :param password: The password needed to log in to the ftp site.
+        :type password: str
+        :param keepalive: Flag to determine whether to keepalive the connection, defaults to False
+        :type keepalive: bool, optional
+        :param debug_lvl: Verbosity level for debugging ftp connection, defaults to 0
+        :type debug_lvl: int, optional
         """
         self._ftpsite = ftpsite
         self._user = user

OrthoEvol/Tools/ftp/ncbiftp.py

@@ -26,10 +26,15 @@ class NcbiFTPClient(BaseFTPClient):
     """Access NCBI's FTP servers with ease."""
 
     def __init__(self, email):
+        """Initialize the NcbiFTPClient.
+
+        :param email: A valid email address to use a password.
+        :type email: str
+        """
         _ncbi = 'ftp.ncbi.nlm.nih.gov'
         super().__init__(_ncbi, user="anonymous", password=email)
-        self._datafmt = '%m-%d-%Y@%I:%M:%S-%p'
-        self._date = str(datetime.now().strftime(self._datafmt))
+        self._datefmt = '%m-%d-%Y@%I:%M:%S-%p'
+        self._date = str(datetime.now().strftime(self._datefmt))
         self.blastpath = '/blast/'
         self.blastdb_path = '/blast/db/'
         self.blastdbv5_path = '/blast/db/v5/'
@@ -38,7 +43,7 @@ def __init__(self, email):
         self.windowmasker_path = self.blastpath + 'windowmasker_files/'
         self.cpus = 1
 
-        # Located in self.blastdb_path
+        # Located in the blastdb path
         self._taxdb = ['taxdb.tar.gz', 'taxdb.tar.gz.md5']
 
         # TODO Turn into a json file, dict, or config
@@ -117,7 +122,11 @@ def download_file(self, filename):
         self.ncbiftp_log.info('%s was downloaded.' % str(filename))
 
     def _download_windowmasker(self, windowmaskerfile):
-        """Download the window masker files."""
+        """Download the window masker files.
+
+        :param windowmaskerfile: A path to a window masker file to download.
+        :type windowmaskerfile: str
+        """
         wm = windowmaskerfile.split(sep='.')
         taxid = wm[0]
         wm_ext = wm[1]
@@ -226,7 +235,20 @@ def getwindowmaskerfiles(self, taxonomy_ids, download_path):
 
     def getrefseqrelease(self, collection_subset, seqtype, seqformat, download_path,
                          extract=True):
-        """Download the refseq release database."""
+        """Download the refseq release database.
+
+        :param collection_subset: [description]
+        :type collection_subset: [type]
+        :param seqtype: [description]
+        :type seqtype: [type]
+        :param seqformat: [description]
+        :type seqformat: [type]
+        :param download_path: [description]
+        :type download_path: [type]
+        :param extract: [description], defaults to True
+        :type extract: bool, optional
+        :raises FileNotFoundError: [description]
+        """
         self.ftp = self._login()
         self.ftp.cwd(self.refseqrelease_path)
         taxon_dirs = self.listdirectories(self.refseqrelease_path)
@@ -240,8 +262,8 @@ def getrefseqrelease(self, collection_subset, seqtype, seqformat, download_path,
         releasefiles = self.listfiles(curpath)
 
         self.files2download = []
-        pattern = re.compile('^' + collection_subset + '[.](.*?)[.]' + seqtype
-                             + '[.]' + seqformat + '[.]gz$')
+        pattern = re.compile('^' + collection_subset +
+                             '[.](.*?)[.]' + seqtype + '[.]' + seqformat + '[.]gz$')
         for releasefile in releasefiles:
             if re.match(pattern, releasefile):
                 self.files2download.append(releasefile)
@@ -276,8 +298,8 @@ def getrefseqrelease(self, collection_subset, seqtype, seqformat, download_path,
     def getblastfasta(self, database_name, download_path, extract=True):
         """Download the fasta sequence database (not formatted).
 
-        :param database_name:
-        :param download_path:
+        :param database_name: The name of the database to download.
+        :param download_path: The path to download the database to.
         :param extract:  (Default value = True)
         """
 
@@ -370,7 +392,6 @@ def getblastdb(self, database_name, download_path, v5=True, extract=True):
 
             except RuntimeError as err:
                 self.ncbiftp_log.error(err)
-                self.cpus = 1
                 # Try to download again
                 self.ncbiftp_log.warning('Attempting to download again.')
                 with ThreadPool(self.cpus) as download_pool: