Merge pull request #134 from jaleezyy/qol_patch

Quality of life changes by including snakemake parameters into main executable

README.md

@@ -47,14 +47,14 @@ If you use this software please [cite](https://doi.org/10.3390/v12080895):
         wget https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86_64.sh
         bash Miniconda3-latest-Linux-x86_64.sh # follow instructions
         source $(conda info --base)/etc/profile.d/conda.sh
-        conda create -n signal -c conda-forge -c bioconda -c defaults snakemake pandas conda
+        conda create -n signal -c conda-forge -c bioconda -c defaults snakemake pandas conda mamba
         conda activate signal
 
 There are some issues with `conda` failing to install newer versions of snakemake
 so alternatively install `mamba` and use that (snakemake has beta support for it within the workflow)
 
         conda install -c conda-forge mamba
-        mamba create -c conda-forge -c bioconda -n signal snakemake pandas conda
+        mamba create -c conda-forge -c bioconda -n signal snakemake pandas conda mamba
         conda activate signal
 
 Additional software dependencies are managed directly by `snakemake` using conda environment files:

scripts/assign_lineages.py

@@ -77,7 +77,7 @@ def update_nextclade_dataset(vers, skip):
 	try:
 		if software_ver != "None": # specific version requested, check if available
 			try:
-				softrequest = subprocess.check_output(f"conda search -c bioconda -f nextclade={software_ver}".split()).decode('utf-8').strip().split()[-3]
+				softrequest = subprocess.check_output(f"conda search -c bioconda -f nextclade", shell=True).split()[-3].strip().decode('utf-8')
 				# check if already installed
 				if softrequest == nextclade_version:
 					print(f"Nextclade {softrequest} already installed! Skipping update!")
@@ -86,15 +86,15 @@ def update_nextclade_dataset(vers, skip):
 					subprocess.run(f"conda install -q -y -c bioconda nextclade={softrequest}", shell=True, check=True)
 			except subprocess.CalledProcessError:
 				print("Cannot find version requested, will ensure latest version!")
-				softrequest = subprocess.check_output(f"conda search -c bioconda -f nextclade".split()).decode('utf-8').strip().split()[-3]
+				softrequest = subprocess.check_output(f"conda search -c bioconda -f nextclade", shell=True).split()[-3].strip().decode('utf-8')
 				# check if already installed
 				if softrequest == nextclade_version:
 					print(f"Nextclade {softrequest} already installed! Skipping update!")
 				else:
 					subprocess.run(f"conda install -q -y -c bioconda nextclade={softrequest}", shell=True, check=True)
 		else:
 			print(f"Installing latest version of Nextclade!")
-			softrequest = subprocess.check_output(f"conda search -c bioconda -f nextclade".split()).decode('utf-8').strip().split()[-3]
+			softrequest = subprocess.check_output(f"conda search -c bioconda -f nextclade", shell=True).split()[-3].strip().decode('utf-8')
 			if softrequest == nextclade_version:
 				print(f"Nextclade {softrequest} already installed! Skipping update!")
 			else:

scripts/signal_postprocess.py

@@ -17,7 +17,7 @@
 
 assert long_git_id.startswith('$Id: ')
 #short_git_id = long_git_id[5:12]
-short_git_id = "v1.5.7"
+short_git_id = "v1.5.8"
 
 # Suppresses matplotlib warning (https://github.com/jaleezyy/covid-19-signal/issues/59)
 # Creates a small memory leak, but it's nontrivial to fix, and won't be a practical concern!

signal.py

@@ -28,6 +28,12 @@ def create_parser():
 	parser.add_argument('--add-breseq', action='store_true', help="Configuration file generator parameter. Set flag to ENABLE optional breseq step (will take more time for analysis to complete)")
 	parser.add_argument('-neg', '--neg-prefix', default=None, help="Configuration file generator parameter. Comma-separated list of negative sontrol sample name(s) or prefix(es). For example, 'Blank' will cover Blank1, Blank2, etc. Recommended if running ncov-tools. Will be left empty, if not provided")
 	parser.add_argument('--dependencies', action='store_true', help="Download data dependencies (under a created 'data' directory) required for SIGNAL analysis and exit. Note: Will override other flags! (~10 GB storage required)")
+	parser.add_argument('-ri', '--rerun-incomplete', action='store_true', help="Snakemake parameter. Re-run any incomplete samples from a previously failed run")
+	parser.add_argument('--unlock', action='store_true', help="Snakemake parameter. Remove a lock on the working directory after a failed run")
+	parser.add_argument('-F', '--forceall', action='store_true', help='Snakemake parameter. Force the re-run of all rules regardless of prior output')
+	parser.add_argument('-n', '--dry-run', action='store_true', help='Snakemake parameter. Do not execute anything and only display what would be done')
+	parser.add_argument('--verbose', action='store_true', help="Snakemake parameter. Display snakemake debugging output")
+	parser.add_argument('-v', '--version', action='store_true', help="Display version number")
 	args, unknown = parser.parse_known_args()
 
 	provided = []
@@ -232,7 +238,12 @@ def write_config_file(run_name, config_file, opt_tasks):
 	# note: add root_dir to determine the root directory of SIGNAL
 	script_path = os.path.join(os.path.abspath(sys.argv[0]).rsplit("/",1)[0])
 	args, allowed = create_parser()
-
+	version = 'v1.5.8'
+	alt_options = []
+
+	if args.version:
+		exit(f"{version}")
+
 	if args.dependencies:
 		print("Downloading necessary reference and dependency files!")
 		download_dependences()
@@ -256,17 +267,25 @@ def write_config_file(run_name, config_file, opt_tasks):
 	if not any([allowed[x] for x in allowed]):
 		exit("No task specified! Please provide at least one of 'all', 'postprocess', or 'ncov_tools'! See 'signal.py -h' for details!")
 	else:
+		if args.verbose: alt_options.append('--verbose')
+		if args.unlock: alt_options.append('--unlock')
+		if args.forceall: alt_options.append('--forceall')
+		if args.dry_run: alt_options.append('--dry-run')
+		if args.rerun_incomplete: alt_options.append('--rerun-incomplete')
+		opt = " ".join(alt_options)
 		for task in allowed:
 			if allowed[task] is True:
 				print(f"Running SIGNAL {task}!")
 				try:
-					subprocess.run(f"snakemake --conda-frontend mamba --configfile {config_file} --cores={args.cores} --use-conda --conda-prefix=$PWD/.snakemake/conda {task} -kp", shell=True, check=True)
+					subprocess.run(f"snakemake --conda-frontend mamba --configfile {config_file} --cores={args.cores} --use-conda --conda-prefix=$PWD/.snakemake/conda {task} -kp {opt}", shell=True, check=True)
 				except subprocess.CalledProcessError: # likely missing mamba 
 					if task == "ncov_tools":
 						check_submodule(os.getcwd())
+					if opt.split(" ")[-1] == '--rerun-incomplete': # remove redundant flag
+						opt = " ".join(opt.split(" ")[:-1])
 					try:
 						print("Retrying...")
-						subprocess.run(f"snakemake --conda-frontend conda --configfile {config_file} --cores={args.cores} --use-conda --conda-prefix=$PWD/.snakemake/conda {task} -kp --rerun-incomplete", shell=True, check=True)
+						subprocess.run(f"snakemake --conda-frontend conda --configfile {config_file} --cores={args.cores} --use-conda --conda-prefix=$PWD/.snakemake/conda {task} -kp --rerun-incomplete {opt}", shell=True, check=True)
 					except subprocess.CalledProcessError:
 						exit(f"Something went wrong running SIGNAL {task}! Check input and try again!")