Patlas recipes

changelog.md

@@ -4,14 +4,28 @@
 
 ### New components
 
-- `maxbin2`: An automatic tool for binning metagenomic sequences
-- `bowtie2`: Align short paired-end sequencing reads to long reference sequences
+- `maxbin2`: An automatic tool for binning metagenomic sequences.
+- `bowtie2`: Align short paired-end sequencing reads to long reference
+sequences.
+
+### New recipes
+
+- `plasmids`: A recipe to perform mapping, mash screen on reads
+and also mash dist for assembly based approaches (all to detect
+plasmdis). This also includes annotation with abricate for the assembly.
+- `plasmids_mapping`: A recipe to perfmorm mapping for plasmids.
+- `plasmids_mash`: A recipe to perform mash screen for plasmids.
+- `plasmids_assembly`: A recipe to perform mash dist for plasmid
+assemblies.
 
 ### Minor/Other changes
 
 - Added "smart" check when the user provides a typo in pipeline string
 for a given process, outputting some "educated" guesses to the the
 terminal.
+- Added "-cr" option to show current recipe `pipeline_string`.
+- Changed the way recipes were being parsed by `proc_collector` for the
+usage of `-l` and `-L` options.
 
 ### Bug fixes
 

flowcraft/flowcraft.py

@@ -15,15 +15,15 @@
     from __init__ import __version__, __build__
     from generator.engine import NextflowGenerator, process_map
     from generator.inspect import NextflowInspector
-    from generator.recipe import brew_recipe
+    from generator.recipe import brew_recipe, available_recipes
     from generator.pipeline_parser import parse_pipeline, SanityError
     from generator.process_details import proc_collector, colored_print
     import generator.error_handling as eh
 except ImportError:
     from flowcraft import __version__, __build__
     from flowcraft.generator.engine import NextflowGenerator, process_map
     from flowcraft.generator.inspect import NextflowInspector
-    from flowcraft.generator.recipe import brew_recipe
+    from flowcraft.generator.recipe import brew_recipe, available_recipes
     from flowcraft.generator.pipeline_parser import parse_pipeline, \
         SanityError
     from flowcraft.generator.process_details import proc_collector, \
@@ -77,6 +77,12 @@ def get_args(args=None):
         "-l", "--short-list", action="store_const", dest="short_list",
         const=True, help="Print a short list of the currently "
                          "available processes")
+    build_parser.add_argument("-cr", "--check-recipe", dest="check_recipe",
+                              action="store_const", const=True,
+                              help="Check tasks that the recipe contain and "
+                                   "their flow. This option might be useful "
+                                   "if a user wants to change some components "
+                                   "of a given recipe, by using the -t option.")
 
     # GENERAL OPTIONS
     parser.add_argument(
@@ -121,14 +127,18 @@ def get_args(args=None):
 
 def validate_build_arguments(args):
 
+    if args.detailed_list or args.short_list:
+        return
+
     if not args.tasks and not args.recipe and not args.check_only \
             and not args.detailed_list and not args.short_list:
         logger.error(colored_print(
             "At least one of these options is required: -t, -r, -c, "
             "-l, -L", "red_bold"))
         sys.exit(1)
 
-    if (args.tasks or args.recipe) and not args.output_nf:
+    if (args.tasks or args.recipe) and not args.check_recipe \
+            and not args.output_nf:
         logger.error(colored_print(
             "Please provide the path and name of the pipeline file using the"
             " -o option.", "red_bold"))
@@ -202,13 +212,26 @@ def build(args):
     # If a recipe is specified, build pipeline based on the
     # appropriate recipe
     if args.recipe:
-        pipeline_string, list_processes = brew_recipe(args)
+        if args.recipe == "innuendo":
+            pipeline_string = brew_recipe(args, available_recipes)
+        else:
+            pipeline_string = available_recipes[args.recipe]
+            if args.tasks:
+                logger.warning(colored_print(
+                    "-t parameter will be ignored for recipe: {}\n"
+                        .format(args.recipe), "yellow_bold")
+                )
+
+        if args.check_recipe:
+            logger.info(colored_print("Pipeline string for recipe: {}"
+                                      .format(args.recipe), "purple_bold"))
+            logger.info(pipeline_string)
+            sys.exit(0)
     else:
         pipeline_string = args.tasks
-        list_processes = None
 
     # used for lists print
-    proc_collector(process_map, args, list_processes)
+    proc_collector(process_map, args, pipeline_string)
 
     logger.info(colored_print("Resulting pipeline string:\n"))
     logger.info(colored_print(pipeline_string + "\n"))

flowcraft/generator/process_details.py

@@ -10,7 +10,8 @@
     "white_bold": "1;38m",
     "white_underline": "4;38m",
     "blue_bold": "1;36m",
-    "purple_bold": "1;34m"
+    "purple_bold": "1;34m",
+    "yellow_bold": "1;93m"
 }
 
 
@@ -75,7 +76,7 @@ def procs_dict_parser(procs_dict):
             ))
 
 
-def proc_collector(process_map, args, processes_list=None):
+def proc_collector(process_map, args, pipeline_string):
     """
     Function that collects all processes available and stores a dictionary of
     the required arguments of each process class to be passed to
@@ -89,9 +90,8 @@ def proc_collector(process_map, args, processes_list=None):
     args: argparse.Namespace
         The arguments passed through argparser that will be access to check the
         type of list to be printed
-    processes_list: list
-        List with all the available processes of a recipe. In case no recipe
-        is passed, the list should come empty.
+    pipeline_string: str
+        the pipeline string
 
     """
 
@@ -120,16 +120,16 @@ def proc_collector(process_map, args, processes_list=None):
         # loops between all process_map Processes
         for name, cls in process_map.items():
 
-            if processes_list:
-                # Skip process if process_list is provided and name not in
-                # processes list
-                if name not in processes_list:
-                    continue
-
             # instantiates each Process class
             cls_inst = cls(template=name)
-            d = {arg_key: vars(cls_inst)[arg_key] for arg_key in vars(cls_inst)
-                 if arg_key in arguments_list}
+
+            # checks if recipe is provided
+            if pipeline_string:
+                if name not in pipeline_string:
+                    continue
+
+            d = {arg_key: vars(cls_inst)[arg_key] for arg_key in
+                 vars(cls_inst) if arg_key in arguments_list}
             procs_dict[name] = d
 
         procs_dict_parser(procs_dict)

flowcraft/generator/recipe.py

@@ -478,8 +478,8 @@ def run_auto_pipeline(self, tasks):
 
         return self.pipeline_string
 
-    def get_process_info(self):
-        return list(self.process_descriptions.keys())
+    # def get_process_info(self):
+    #     return list(self.process_descriptions.keys())
 
 
 class Innuendo(Recipe):
@@ -512,17 +512,12 @@ def __init__(self, *args, **kwargs):
             "mlst": [False, "pilon", "abricate|prokka|chewbbaca|sistr"],
             "sistr": [True, "mlst", None],
             "abricate": [True, "mlst", None],
-            "prokka": [True, "mlst", None],
+            #"prokka": [True, "mlst", None],
             "chewbbaca": [True, "mlst", None]
         }
 
 
-available_recipes = {
-    "innuendo": Innuendo
-}
-
-
-def brew_recipe(args):
+def brew_recipe(args, available_recipes):
     """Brews a given list of processes according to the recipe
 
     Parameters
@@ -555,13 +550,23 @@ def brew_recipe(args):
     else:
         input_processes = args.tasks
 
-    # Get the list of processes for that recipe
-    list_processes = automatic_pipeline.get_process_info()
     # Validate the provided pipeline processes
     validated = automatic_pipeline.validate_pipeline(input_processes)
     if not validated:
         sys.exit(1)
     # Get the final pipeline string
     pipeline_string = automatic_pipeline.run_auto_pipeline(input_processes)
 
-    return pipeline_string, list_processes
+    return pipeline_string
+
+
+# A dictionary of quick recipes
+available_recipes = {
+    "innuendo": Innuendo,
+    "plasmids": "integrity_coverage fastqc_trimmomatic (spades pilon "
+              "(mash_dist | abricate) | mash_screen | mapping_patlas)",
+    "plasmids_mapping": "integrity_coverage fastqc_trimmomatic mapping_patlas",
+    "plasmids_assembly": "integrity_coverage fastqc_trimmomatic (spades pilon"
+                         " mash_dist)",
+    "plasmids_mash": "integrity_coverage fastqc_trimmomatic mash_screen",
+}

flowcraft/tests/test_assemblerflow.py

@@ -14,24 +14,6 @@ def tmp():
     shutil.rmtree("temp")
 
 
-def test_list_short():
-
-    sys.argv.append(1)
-    args = af.get_args(["build", "-l"])
-
-    with pytest.raises(SystemExit):
-        af.build(args)
-
-
-def test_list_long():
-
-    sys.argv.append(1)
-    args = af.get_args(["build", "-L"])
-
-    with pytest.raises(SystemExit):
-        af.build(args)
-
-
 def test_check():
 
     sys.argv.append(1)

flowcraft/tests/test_process_details.py

@@ -19,14 +19,18 @@ def test_long_list():
 
     arguments = af.get_args(["build", "-L"])
 
+    pipeline_string = "fastqc trimmomatic"
+
     with pytest.raises(SystemExit):
-        pd.proc_collector(process_map, arguments)
+        pd.proc_collector(process_map, arguments, pipeline_string)
 
 
 def test_short_list():
 
     arguments = af.get_args(["build", "-l"])
 
+    pipeline_string = "fastqc trimmomatic"
+
     with pytest.raises(SystemExit):
-        pd.proc_collector(process_map, arguments)
+        pd.proc_collector(process_map, arguments, pipeline_string)