Merge pull request #44 from vinisalazar/dev

Merge v0.1.21

CHANGELOG.md

@@ -8,6 +8,15 @@
     * .write_paths_to_file, .copy_files_to_dir(), .link_files_to_dir() [ ]
 * Add logger calls when saving to JSON and uploading to ProvStore [ ]
 
+### v0.1.21
+* Add _config argument to `bp.load_project()` (this is a temporary fix) [x]
+* Add db property to Config class (prevents bug when setting DB path) [x]
+* Improve a few docstrings [x]
+* Refactor sha1 as sha256 [x]  
+* Add add and radd dunder methods for Directory and File [x]  
+* Package workflows in single module [x]
+* Add FastTree PresetProgram [x]
+
 ### v0.1.20
 * Debug graphical DOT output [x]
 * Add Muscle PresetProgram [x]

bioprov/__init__.py

@@ -2,7 +2,7 @@
 __license__ = "MIT"
 __maintainer__ = "Vini Salazar"
 __url__ = "https://github.com/vinisalazar/bioprov"
-__version__ = "0.1.20"
+__version__ = "0.1.21"
 
 
 """
@@ -11,7 +11,7 @@
 Inherits objects from the src/ package.
 """
 
-from .src.config import config, Environment, BioProvDB
+from .src.config import config, Environment, BioProvDB, Config
 from .src.files import File, SeqFile, Directory
 from .src.main import (
     Program,

bioprov/bioprov

@@ -3,7 +3,7 @@ __author__ = "Vini Salazar"
 __license__ = "MIT"
 __maintainer__ = "Vini Salazar"
 __url__ = "https://github.com/vinisalazar/bioprov"
-__version__ = "0.1.20"
+__version__ = "0.1.21"
 
 """
 Executable that goes in $PATH. Code for the command-line is on the bioprov.py module.

bioprov/bioprov.py

@@ -2,7 +2,7 @@
 __license__ = "MIT"
 __maintainer__ = "Vini Salazar"
 __url__ = "https://github.com/vinisalazar/bioprov"
-__version__ = "0.1.20"
+__version__ = "0.1.21"
 
 """
 BioProv command-line application. This module holds the main executable.

bioprov/data/__init__.py

@@ -2,7 +2,7 @@
 __license__ = "MIT"
 __maintainer__ = "Vini Salazar"
 __url__ = "https://github.com/vinisalazar/bioprov"
-__version__ = "0.1.20"
+__version__ = "0.1.21"
 
 
 """

bioprov/programs/__init__.py

@@ -2,7 +2,7 @@
 __license__ = "MIT"
 __maintainer__ = "Vini Salazar"
 __url__ = "https://github.com/vinisalazar/bioprov"
-__version__ = "0.1.20"
+__version__ = "0.1.21"
 
 
 from .programs import (
@@ -12,6 +12,7 @@
     blastp,
     muscle,
     mafft,
+    fasttree,
     prokka,
     kallisto_quant,
     kaiju,

bioprov/programs/programs.py

@@ -2,7 +2,7 @@
 __license__ = "MIT"
 __maintainer__ = "Vini Salazar"
 __url__ = "https://github.com/vinisalazar/bioprov"
-__version__ = "0.1.20"
+__version__ = "0.1.21"
 
 
 """
@@ -179,6 +179,26 @@ def mafft(sample, input_tag="input", extra_flags=None):
     return _mafft
 
 
+def fasttree(sample, input_tag="input", extra_flags=None):
+    """
+    :param Sample sample: Instance of BioProv.Sample.
+    :param str input_tag:  A tag for the input multifasta file.
+    :param list extra_flags: A list of extra parameters to pass to FastTree.
+    :return: Instance of PresetProgram containing FastTree.
+    :rtype: BioProv.PresetProgram.
+    """
+    _fasttree = PresetProgram(
+        name="fasttree",
+        sample=sample,
+        input_files={"": input_tag},
+        output_files={">": ("tree", ".tree")},
+        preffix_tag=input_tag,
+        extra_flags=extra_flags,
+    )
+
+    return _fasttree
+
+
 def kallisto_quant(sample, index, output_dir="./", extra_flags=None):
     """
     Run kallisto's alignment and quantification

bioprov/src/__init__.py

@@ -2,7 +2,7 @@
 __license__ = "MIT"
 __maintainer__ = "Vini Salazar"
 __url__ = "https://github.com/vinisalazar/bioprov"
-__version__ = "0.1.20"
+__version__ = "0.1.21"
 
 
 """

bioprov/src/config.py

@@ -2,7 +2,7 @@
 __license__ = "MIT"
 __maintainer__ = "Vini Salazar"
 __url__ = "https://github.com/vinisalazar/bioprov"
-__version__ = "0.1.20"
+__version__ = "0.1.21"
 
 
 """
@@ -20,7 +20,7 @@
 
 from bioprov import __file__ as bp_file
 from bioprov.data import data_dir, genomes_dir
-from bioprov.utils import serializer, dict_to_sha1, serializer_filter, create_logger
+from bioprov.utils import serializer, dict_to_sha256, serializer_filter, create_logger
 
 
 class Config:
@@ -36,19 +36,18 @@ def __init__(self, db_path=None, threads=0):
         # This duplication is to order the keys in the __dict__ attribute.
         self.user = None
         self.env = Environment()
-        self.user = self.env.user
+        if self.user is None:
+            self.user = self.env.user
         if not threads:
             threads = int(os.cpu_count() / 2)
-        self.db = None
-        self.db_path = None
         self.threads = threads
         self.bioprov_dir = Path(bp_file).parent
         self.data = data_dir
         self.genomes = genomes_dir
         if db_path is None:
             db_path = self.bioprov_dir.joinpath("db.json")
-        self.db_path = db_path
-        self.db = BioProvDB(self.db_path)
+        self._db_path = db_path
+        self._db = BioProvDB(self.db_path)
         self._provstore_file = None
         self._provstore_user = None
         self._provstore_token = None
@@ -59,6 +58,18 @@ def __init__(self, db_path=None, threads=0):
     def __repr__(self):
         return f"BioProv Config class set in {__file__}"
 
+    @property
+    def db(self):
+        return self._db
+
+    @property
+    def db_path(self):
+        return self._db_path
+
+    @db_path.setter
+    def db_path(self, value):
+        self._db_path = value
+
     def db_all(self):
         """
         :return: List all items in BioProv database.
@@ -273,7 +284,7 @@ def update(self):
         :return: Sets attributes to self.
         """
         env_dict = dict(os.environ.items())
-        env_hash = dict_to_sha1(env_dict)
+        env_hash = dict_to_sha256(env_dict)
         if env_hash != self.env_hash:
             self.env_dict = env_dict
             self.env_hash = env_hash

bioprov/src/files.py

@@ -2,7 +2,7 @@
 __license__ = "MIT"
 __maintainer__ = "Vini Salazar"
 __url__ = "https://github.com/vinisalazar/bioprov"
-__version__ = "0.1.20"
+__version__ = "0.1.21"
 
 
 """
@@ -21,7 +21,7 @@
     Warnings,
     serializer_filter,
     serializer,
-    file_to_sha1,
+    file_to_sha256,
     pattern_replacer,
 )
 
@@ -54,7 +54,7 @@ def __init__(self, path, tag=None, attributes=None, _get_hash=True):
         self._exists = self.path.exists()
         self.size = get_size(self.path)
         self.raw_size = get_size(self.path, convert=False)
-        self._sha1 = file_to_sha1(self.path)
+        self._sha256 = file_to_sha256(self.path)
 
         # Provenance attributes
         self._entity = None
@@ -65,14 +65,20 @@ def __repr__(self):
     def __str__(self):
         return self.__repr__()
 
+    def __add__(self, other):
+        return str(self) + other
+
+    def __radd__(self, other):
+        return other + str(self)
+
     @property
-    def sha1(self):
-        self._sha1 = file_to_sha1(self.path)
-        return self._sha1
+    def sha256(self):
+        self._sha256 = file_to_sha256(self.path)
+        return self._sha256
 
-    @sha1.setter
-    def sha1(self, value):
-        self._sha1 = value  # no cover
+    @sha256.setter
+    def sha256(self, value):
+        self._sha256 = value  # no cover
 
     @property
     def exists(self):
@@ -107,21 +113,20 @@ def replace_path(self, old_terms, new, warnings=False):
 
         :param old_terms: Terms to be replaced in the path.
         :param new: New term.
-        :param warnings: Whether to warn if sha1 checksum differs or file does not exist.
+        :param warnings: Whether to warn if sha256 checksum differs or file does not exist.
 
         :return: Updates self.
         """
-        old_hash, old_exists = self._sha1, self._exists
+        old_hash, old_exists = self._sha256, self._exists
         self.path = Path(pattern_replacer(str(self.path), old_terms, new))
-        # TODO: replace these print statements for logger warning/debug level
         if warnings:
             if not self.exists and old_exists:
                 logging.warning(
                     f"File {self.path} was marked as existing but was not found."
                 )
-            if old_hash and self.sha1 != old_hash and self.exists:  # no cover
+            if old_hash and self.sha256 != old_hash and self.exists:  # no cover
                 logging.warning(
-                    f"File {self.path} previous sha1 checksum differs from the current."
+                    f"File {self.path} previous sha256 checksum differs from the current."
                 )
 
     def serializer(self):
@@ -153,7 +158,7 @@ def replace_path(self, old_terms, new, warnings=False):
 
         :param old_terms: Terms to be replaced in the path.
         :param new: New term.
-        :param warnings: Whether to warn if sha1 checksum differs or file does not exist.
+        :param warnings: Whether to warn if sha256 checksum differs or file does not exist.
 
         :return: Updates self.
         """
@@ -172,6 +177,12 @@ def __repr__(self):
     def __str__(self):
         return self.__repr__()
 
+    def __add__(self, other):
+        return str(self) + other
+
+    def __radd__(self, other):
+        return other + str(self)
+
     @property
     def exists(self):
         self._exists = self.path.exists()
@@ -433,7 +444,7 @@ def seqrecordgenerator(path, format, parser="seq", warnings=False):
     :param path: Path to file.
     :param format: format to pass to SeqIO.parse().
     :param parser: Whether to import records with SeqIO (default) or AlignIO
-    :param warnings: Whether to warn if sha1 checksum differs or file does not exist.
+    :param warnings: Whether to warn if sha256 checksum differs or file does not exist.
 
     :return: A generator of SeqRecords.
     """