Merge pull request #184 from rhysnewell/doco_updates

Doco updates

README.md

@@ -57,7 +57,12 @@ conda env create -n aviary -f aviary.yml
 conda activate aviary
 pip install -e .
 ```
+The `aviary` executable can then be run from any directory. Since the code in
+this directory is then used for running, any updates made there will be
+immediately available. We recommend this mode for developing and debugging
+aviary.
 
+## Checking installation
 Whatever option you choose, running `aviary --help` should return the following
 output:
 
@@ -86,22 +91,6 @@ Utility modules:
 
 ```
 
-Upon first running aviary you will be prompted to input the location for where you would like
-your conda environments to be stored, the GTDB release installed on your system, the location of your
-EnrichM database, and the location of your BUSCO database. These locations will be stored as environment
-variables, but for aviary to be able to use those environment variables you will have to either source your .bashrc
-or reactivate your conda environment depending on whether you installed aviary within a conda environment or not:
-
-```
-conda deactivate; conda activate aviary
-
-OR
-
-source ~/.bashrc
-```
-
-These environment variables can be reset using `aviary configure`
-
 ## Databases
 
 Aviary uses programs which require access to locally stored databases. 
@@ -111,7 +100,7 @@ The **required** databases are as follows:
 * [GTDB](https://gtdb.ecogenomic.org/downloads)
 * [EggNog](https://github.com/eggnogdb/eggnog-mapper/wiki/eggNOG-mapper-v2.1.5-to-v2.1.8#setup)
 * [CheckM2](https://github.com/chklovski/CheckM2)
-
+* [SingleM](https://wwood.github.io/singlem/)
 
 ### Installing databases
 

aviary/aviary.py

@@ -100,6 +100,17 @@ def str2bool(v):
     else:
         raise argparse.ArgumentTypeError('Boolean value expected.')
 
+def add_workflow_arg(parser, default, help=None):
+    if help is None:
+        help = 'Main workflow to run. This is the snakemake target rule to run.'
+    parser.add_argument(
+        '-w', '--workflow',
+        help=help,
+        dest='workflow',
+        nargs="+",
+        default=default,
+    )
+
 def main():
     if len(sys.argv) == 1 or sys.argv[1] == '-h' or sys.argv[1] == '--help':
         phelp()
@@ -832,13 +843,7 @@ def main():
 
 
 
-    assemble_options.add_argument(
-        '-w', '--workflow',
-        help='Main workflow to run',
-        dest='workflow',
-        nargs="+",
-        default=['complete_assembly_with_qc'],
-    )
+    add_workflow_arg(assemble_options, ['complete_assembly_with_qc'])
 
     ##########################  ~ RECOVER ~   ###########################
 
@@ -862,13 +867,7 @@ def main():
         required=False,
     )
 
-    recover_options.add_argument(
-        '-w', '--workflow',
-        help='Main workflow to run',
-        dest='workflow',
-        nargs="+",
-        default=['recover_mags'],
-    )
+    add_workflow_arg(recover_options, ['recover_mags'])
 
     recover_options.add_argument(
         '--perform-strain-analysis', '--perform_strain_analysis',
@@ -902,13 +901,7 @@ def main():
         required=False,
     )
 
-    annotate_options.add_argument(
-        '-w', '--workflow',
-        help='Main workflow to run',
-        dest='workflow',
-        nargs="+",
-        default=['annotate'],
-    )
+    add_workflow_arg(annotate_options, ['annotate'])
 
     ##########################  ~ diversity ~   ###########################
 
@@ -933,13 +926,7 @@ def main():
         required=False,
     )
 
-    diversity_options.add_argument(
-        '-w', '--workflow',
-        help='Main workflow to run',
-        dest='workflow',
-        nargs="+",
-        default=['lorikeet'],
-    )
+    add_workflow_arg(diversity_options, ['lorikeet'])
 
     diversity_options.add_argument(
         '--perform-strain-analysis', '--perform_strain_analysis',
@@ -975,13 +962,7 @@ def main():
         required=True,
     )
 
-    cluster_options.add_argument(
-        '-w', '--workflow',
-        help='Main workflow to run',
-        dest='workflow',
-        nargs="+",
-        default=['complete_cluster'],
-    )
+    add_workflow_arg(cluster_options, ['complete_cluster'])
 
     ##########################  ~ VIRAL ~   ###########################
 
@@ -997,13 +978,7 @@ def main():
  
                                           ''')
 
-    viral_options.add_argument(
-        '-w', '--workflow',
-        help='Main workflow to run',
-        dest='workflow',
-        nargs="+",
-        default=['create_webpage_genotype'],
-    )
+    add_workflow_arg(viral_options, ['create_webpage_genotype'])
 
     ##########################   ~ COMPLETE ~  ###########################
 
@@ -1028,13 +1003,7 @@ def main():
         required=False,
     )
 
-    complete_options.add_argument(
-        '-w', '--workflow',
-        help='Main workflow to run',
-        dest='workflow',
-        nargs="+",
-        default=['get_bam_indices', 'recover_mags', 'annotate', 'lorikeet'],
-    )
+    add_workflow_arg(complete_options, ['get_bam_indices', 'recover_mags', 'annotate', 'lorikeet'])
 
     ##########################  ~ ISOLATE ~  ###########################
 
@@ -1050,13 +1019,7 @@ def main():
                                  
                                              ''')
 
-    isolate_options.add_argument(
-        '-w', '--workflow',
-        help='Main workflows to run',
-        dest='workflow',
-        nargs="+",
-        default=['circlator'],
-    )
+    add_workflow_arg(isolate_options, ['circlator'])
 
     ##########################   ~ BATCH ~  ###########################
 
@@ -1131,12 +1094,10 @@ def main():
         default='95'
     )
 
-    batch_options.add_argument(
-        '-w', '--workflow',
-        help='Main workflow to run for each sample',
-        dest='workflow',
-        nargs="+",
-        default=['get_bam_indices', 'recover_mags', 'annotate', 'lorikeet'],
+    add_workflow_arg(
+        batch_options,
+        ['get_bam_indices', 'recover_mags', 'annotate', 'lorikeet'],
+        help='Main workflow (snakemake target rule) to run for each sample'
     )
 
     ##########################   ~ configure ~  ###########################
@@ -1188,13 +1149,7 @@ def main():
         required=False,
     )
 
-    configure_options.add_argument(
-        '-w', '--workflow',
-        help=argparse.SUPPRESS,
-        dest='workflow',
-        nargs="+",
-        default=['download_databases'],
-    )
+    add_workflow_arg(configure_options, ['download_databases'], help=argparse.SUPPRESS)
 
     ###########################################################################
     # ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ #
@@ -1303,6 +1258,8 @@ def manage_env_vars(args):
             args.eggnog_db_path = Config.get_software_db_path('EGGNOG_DATA_DIR', '--eggnog-db-path')
         if args.checkm2_db_path is None:
             args.checkm2_db_path = Config.get_software_db_path('CHECKM2DB', '--checkm2-db-path')
+        if args.singlem_metapackage_path is None:
+            args.singlem_db_path = Config.get_software_db_path('SINGLEM_METAPACKAGE_PATH', '--singlem-metapackage-path')
     except AttributeError:
         pass