From b5db20016556c225086c0bcc8b87dd0a319457db Mon Sep 17 00:00:00 2001
From: burri0000 <dominik.burri@unibas.ch>
Date: Mon, 12 Jul 2021 13:58:52 +0200
Subject: [PATCH 01/11] add author information (#78)
---
tests/input_files/config.mutliple_lanes.yml | 2 ++
tests/input_files/config.yaml | 2 ++
tests/input_files/config_alfa.yaml | 2 ++
workflow/Snakefile | 4 ++++
4 files changed, 10 insertions(+)
diff --git a/tests/input_files/config.mutliple_lanes.yml b/tests/input_files/config.mutliple_lanes.yml
index 3d04a01..ada2c6b 100644
--- a/tests/input_files/config.mutliple_lanes.yml
+++ b/tests/input_files/config.mutliple_lanes.yml
@@ -10,4 +10,6 @@
report_description: "No description provided by user"
report_logo: "../../images/logo.128px.png"
report_url: "https://zavolan.biozentrum.unibas.ch/"
+ author_name: "NA"
+ author_email: "NA"
...
diff --git a/tests/input_files/config.yaml b/tests/input_files/config.yaml
index 853a8a7..f7a7410 100644
--- a/tests/input_files/config.yaml
+++ b/tests/input_files/config.yaml
@@ -10,4 +10,6 @@
report_description: "No description provided by user"
report_logo: "../../images/logo.128px.png"
report_url: "https://zavolan.biozentrum.unibas.ch/"
+ author_name: "NA"
+ author_email: "NA"
...
diff --git a/tests/input_files/config_alfa.yaml b/tests/input_files/config_alfa.yaml
index cdcfb8b..60c1e0a 100644
--- a/tests/input_files/config_alfa.yaml
+++ b/tests/input_files/config_alfa.yaml
@@ -9,4 +9,6 @@
report_description: "No description provided by user"
report_logo: "../../images/logo.128px.png"
report_url: "https://zavolan.biozentrum.unibas.ch/"
+ author_name: "NA"
+ author_email: "NA"
...
diff --git a/workflow/Snakefile b/workflow/Snakefile
index d46c511..5f2434e 100644
--- a/workflow/Snakefile
+++ b/workflow/Snakefile
@@ -1547,6 +1547,8 @@ rule prepare_multiqc_config:
logo_path = config['report_logo'],
multiqc_intro_text = config['report_description'],
url = config['report_url'],
+ author_name = config['author_name'],
+ author_email = config['author_email'],
additional_params = parse_rule_config(
rule_config,
current_rule=current_rule,
@@ -1572,6 +1574,8 @@ rule prepare_multiqc_config:
--intro-text '{params.multiqc_intro_text}' \
--custom-logo {params.logo_path} \
--url '{params.url}' \
+ --author-name {params.author_name} \
+ --author-email {params.author_email} \
{params.additional_params}) \
1> {log.stdout} 2> {log.stderr}"
--
GitLab
From ef68b7658b99c69e52f8a51736e0af03e2e969db Mon Sep 17 00:00:00 2001
From: burri0000 <dominik.burri@unibas.ch>
Date: Tue, 13 Jul 2021 11:58:31 +0200
Subject: [PATCH 02/11] validate config on required and optional fields #174
---
tests/input_files/config.yaml | 13 ++++++-----
workflow/Snakefile | 41 ++++++++++++++++++++++++++++++-----
2 files changed, 44 insertions(+), 10 deletions(-)
diff --git a/tests/input_files/config.yaml b/tests/input_files/config.yaml
index f7a7410..f3cb261 100644
--- a/tests/input_files/config.yaml
+++ b/tests/input_files/config.yaml
@@ -1,4 +1,5 @@
---
+ # Required fields
samples: "../input_files/samples.tsv"
rule_config: "../input_files/rule_config.yaml"
output_dir: "results"
@@ -7,9 +8,11 @@
salmon_indexes: "results/salmon_indexes"
star_indexes: "results/star_indexes"
alfa_indexes: "results/alfa_indexes"
- report_description: "No description provided by user"
- report_logo: "../../images/logo.128px.png"
- report_url: "https://zavolan.biozentrum.unibas.ch/"
- author_name: "NA"
- author_email: "NA"
+ # Optional fields
+ optional:
+ report_description: "No description provided by user"
+ report_logo: "../../images/logo.128px.png"
+ report_url: "https://zavolan.biozentrum.unibas.ch/"
+ author_name: "NA"
+ author_email: "NA"
...
diff --git a/workflow/Snakefile b/workflow/Snakefile
index 5f2434e..b5e295d 100644
--- a/workflow/Snakefile
+++ b/workflow/Snakefile
@@ -17,6 +17,37 @@ samples_table = pd.read_csv(
sep="\t",
)
+# Validate config with template config
+template_config_path = os.path.join(
+ workflow.current_basedir,
+ '..',
+ 'tests',
+ 'input_files',
+ 'config.yaml')
+try:
+ with open(template_config_path) as _file:
+ template_config = yaml.safe_load(_file)
+ logger.info(f"Loaded template config from {template_config_path}.")
+ # Check if config contains required fields in example config
+ for key in template_config:
+ if key != 'optional':
+ assert key in config
+except FileNotFoundError:
+ logger.error(f"No config file found at {template_config_path}! ")
+ raise
+except AssertionError:
+ logger.error(f"Required config fields from {template_config_path} do not match {config.keys()}")
+ raise
+# Check if optional field available
+if 'optional' not in config:
+ config['optional'] = {}
+# Check optional fields and include in config if not present
+for optkey, value in template_config['optional'].items():
+ if optkey not in config['optional']:
+ config['optional'][optkey] = value
+ logger.info(f"Added optional field: \"{optkey}\" with value: \"{value}\" to config.")
+
+
# Parse YAML rule config file
if 'rule_config' in config and config['rule_config']:
try:
@@ -1544,11 +1575,11 @@ rule prepare_multiqc_config:
"multiqc_config.yaml")
params:
- logo_path = config['report_logo'],
- multiqc_intro_text = config['report_description'],
- url = config['report_url'],
- author_name = config['author_name'],
- author_email = config['author_email'],
+ logo_path = config['optional']['report_logo'],
+ multiqc_intro_text = config['optional']['report_description'],
+ url = config['optional']['report_url'],
+ author_name = config['optional']['author_name'],
+ author_email = config['optional']['author_email'],
additional_params = parse_rule_config(
rule_config,
current_rule=current_rule,
--
GitLab
From 1e5e058dc91f1d2dbe1da88e5f5f266fe3eec771 Mon Sep 17 00:00:00 2001
From: burri0000 <dominik.burri@unibas.ch>
Date: Tue, 13 Jul 2021 14:31:43 +0200
Subject: [PATCH 03/11] adjust all config files and test cases to new optional
field.
---
scripts/prepare_inputs.py | 7 +++---
tests/input_files/config.mutliple_lanes.yml | 11 +++++-----
tests/input_files/config.yaml | 2 +-
tests/input_files/config_alfa.yaml | 22 +++++++++----------
.../expected_output.md5 | 2 +-
workflow/Snakefile | 8 +++----
6 files changed, 26 insertions(+), 26 deletions(-)
diff --git a/scripts/prepare_inputs.py b/scripts/prepare_inputs.py
index 426902a..e47fa3d 100755
--- a/scripts/prepare_inputs.py
+++ b/scripts/prepare_inputs.py
@@ -678,9 +678,10 @@ def main(args):
salmon_indexes: "{salmon_indexes}"
star_indexes: "{star_indexes}"
alfa_indexes: "{alfa_indexes}"
- report_description: "{args.description}"
- report_logo: "{args.logo}"
- report_url: "{args.url}"
+ optional:
+ report_description: "{args.description}"
+ report_logo: "{args.logo}"
+ report_url: "{args.url}"
...
'''
args.config_file.write(config_file_content)
diff --git a/tests/input_files/config.mutliple_lanes.yml b/tests/input_files/config.mutliple_lanes.yml
index ada2c6b..016692d 100644
--- a/tests/input_files/config.mutliple_lanes.yml
+++ b/tests/input_files/config.mutliple_lanes.yml
@@ -1,15 +1,14 @@
---
samples: "../input_files/samples.multiple_lanes.tsv"
- rule_config: "../input_files/rule_config.yaml"
output_dir: "results"
log_dir: "logs"
kallisto_indexes: "results/kallisto_indexes"
salmon_indexes: "results/salmon_indexes"
star_indexes: "results/star_indexes"
alfa_indexes: "results/alfa_indexes"
- report_description: "No description provided by user"
- report_logo: "../../images/logo.128px.png"
- report_url: "https://zavolan.biozentrum.unibas.ch/"
- author_name: "NA"
- author_email: "NA"
+ optional:
+ rule_config: "../input_files/rule_config.yaml"
+ report_description: "No description provided by user"
+ report_logo: "../../images/logo.128px.png"
+ report_url: "https://zavolan.biozentrum.unibas.ch/"
...
diff --git a/tests/input_files/config.yaml b/tests/input_files/config.yaml
index f3cb261..54edf9c 100644
--- a/tests/input_files/config.yaml
+++ b/tests/input_files/config.yaml
@@ -1,7 +1,6 @@
---
# Required fields
samples: "../input_files/samples.tsv"
- rule_config: "../input_files/rule_config.yaml"
output_dir: "results"
log_dir: "logs"
kallisto_indexes: "results/kallisto_indexes"
@@ -10,6 +9,7 @@
alfa_indexes: "results/alfa_indexes"
# Optional fields
optional:
+ rule_config: "../input_files/rule_config.yaml"
report_description: "No description provided by user"
report_logo: "../../images/logo.128px.png"
report_url: "https://zavolan.biozentrum.unibas.ch/"
diff --git a/tests/input_files/config_alfa.yaml b/tests/input_files/config_alfa.yaml
index 60c1e0a..06decc6 100644
--- a/tests/input_files/config_alfa.yaml
+++ b/tests/input_files/config_alfa.yaml
@@ -1,14 +1,14 @@
---
samples: "../input_files/samples_alfa.tsv"
- output_dir: "results/"
- log_dir: "logs/"
- kallisto_indexes: "results/kallisto_indexes/"
- salmon_indexes: "results/salmon_indexes/"
- star_indexes: "results/star_indexes/"
- alfa_indexes: "results/alfa_indexes/"
- report_description: "No description provided by user"
- report_logo: "../../images/logo.128px.png"
- report_url: "https://zavolan.biozentrum.unibas.ch/"
- author_name: "NA"
- author_email: "NA"
+ output_dir: "results"
+ log_dir: "logs"
+ kallisto_indexes: "results/kallisto_indexes"
+ salmon_indexes: "results/salmon_indexes"
+ star_indexes: "results/star_indexes"
+ alfa_indexes: "results/alfa_indexes"
+ optional:
+ rule_config: "../input_files/rule_config.yaml"
+ report_description: "No description provided by user"
+ report_logo: "../../images/logo.128px.png"
+ report_url: "https://zavolan.biozentrum.unibas.ch/"
...
diff --git a/tests/test_scripts_prepare_inputs_table/expected_output.md5 b/tests/test_scripts_prepare_inputs_table/expected_output.md5
index 53afed5..dc555cc 100644
--- a/tests/test_scripts_prepare_inputs_table/expected_output.md5
+++ b/tests/test_scripts_prepare_inputs_table/expected_output.md5
@@ -1,2 +1,2 @@
-40bd0f0fcecdd0d9bc932f63c2811478 config.yaml
+90e90bb60335a2e17cd40cbf058af7b7 config.yaml
c8dcc5a203e9046806c4090525960151 samples.tsv
diff --git a/workflow/Snakefile b/workflow/Snakefile
index b5e295d..e5d270b 100644
--- a/workflow/Snakefile
+++ b/workflow/Snakefile
@@ -49,13 +49,13 @@ for optkey, value in template_config['optional'].items():
# Parse YAML rule config file
-if 'rule_config' in config and config['rule_config']:
+if 'rule_config' in config['optional'] and config['optional']['rule_config']:
try:
- with open(config['rule_config']) as _file:
+ with open(config['optional']['rule_config']) as _file:
rule_config = yaml.safe_load(_file)
- logger.info(f"Loaded rule_config from {config['rule_config']}.")
+ logger.info(f"Loaded rule_config from {config['optional']['rule_config']}.")
except FileNotFoundError:
- logger.error(f"No rule config file found at {config['rule_config']}. Either provide file or remove rule_config parameter from config.yaml! ")
+ logger.error(f"No rule config file found at {config['optional']['rule_config']}. Either provide file or remove rule_config parameter from config.yaml! ")
raise
else:
rule_config = {}
--
GitLab
From efa866595c1aec6a089a8f1e2efd57e142de3da4 Mon Sep 17 00:00:00 2001
From: burri0000 <dominik.burri@unibas.ch>
Date: Wed, 14 Jul 2021 09:59:24 +0200
Subject: [PATCH 04/11] Report missing required keys, restructure rule config
handling.
---
workflow/Snakefile | 37 +++++++++++++++++++------------------
1 file changed, 19 insertions(+), 18 deletions(-)
diff --git a/workflow/Snakefile b/workflow/Snakefile
index e5d270b..346f081 100644
--- a/workflow/Snakefile
+++ b/workflow/Snakefile
@@ -28,36 +28,37 @@ try:
with open(template_config_path) as _file:
template_config = yaml.safe_load(_file)
logger.info(f"Loaded template config from {template_config_path}.")
- # Check if config contains required fields in example config
- for key in template_config:
- if key != 'optional':
- assert key in config
except FileNotFoundError:
logger.error(f"No config file found at {template_config_path}! ")
raise
-except AssertionError:
- logger.error(f"Required config fields from {template_config_path} do not match {config.keys()}")
- raise
+# Check if config contains required fields in example config
+missing = [key for key in template_config if key not in config and key != 'optional']
+if missing:
+ err_msg = f"Required keys missing: {missing}"
+ logger.error(err_msg)
+ raise ValueError(err_msg)
# Check if optional field available
-if 'optional' not in config:
+if 'optional' not in config or config['optional'] != dict:
+ logger.info(f'No "optional" field found or no valid configuration for config.yaml.')
config['optional'] = {}
# Check optional fields and include in config if not present
for optkey, value in template_config['optional'].items():
if optkey not in config['optional']:
config['optional'][optkey] = value
- logger.info(f"Added optional field: \"{optkey}\" with value: \"{value}\" to config.")
+ logger.info(f'Set default value for optional parameter "{optkey}" to : "{value}"')
# Parse YAML rule config file
-if 'rule_config' in config['optional'] and config['optional']['rule_config']:
- try:
- with open(config['optional']['rule_config']) as _file:
- rule_config = yaml.safe_load(_file)
- logger.info(f"Loaded rule_config from {config['optional']['rule_config']}.")
- except FileNotFoundError:
- logger.error(f"No rule config file found at {config['optional']['rule_config']}. Either provide file or remove rule_config parameter from config.yaml! ")
- raise
-else:
+try:
+ with open(config['optional']['rule_config']) as _file:
+ rule_config = yaml.safe_load(_file)
+ logger.info(f"Loaded rule_config from {config['optional']['rule_config']}.")
+except TypeError:
+ logger.error(f'No string supplied at field "rule_config", but: {config["optional"]["rule_config"]}')
+except FileNotFoundError:
+ logger.error(f"No rule config file found at {config['optional']['rule_config']}. Either provide file or remove rule_config parameter from config.yaml! ")
+ raise
+except KeyError:
rule_config = {}
logger.warning(f"No rule config specified: using default values for all tools.")
--
GitLab
From 3baa76e21f951eb7b8e33036bb2c62bbcc2cbe78 Mon Sep 17 00:00:00 2001
From: burri0000 <dominik.burri@unibas.ch>
Date: Wed, 14 Jul 2021 10:34:53 +0200
Subject: [PATCH 05/11] correct error to create empty optional dict.
---
workflow/Snakefile | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/workflow/Snakefile b/workflow/Snakefile
index 346f081..8ffa423 100644
--- a/workflow/Snakefile
+++ b/workflow/Snakefile
@@ -38,8 +38,8 @@ if missing:
logger.error(err_msg)
raise ValueError(err_msg)
# Check if optional field available
-if 'optional' not in config or config['optional'] != dict:
- logger.info(f'No "optional" field found or no valid configuration for config.yaml.')
+if 'optional' not in config or type(config['optional']) not in [OrderedDict, dict]:
+ logger.info(f'No "optional" field found or no valid configuration.')
config['optional'] = {}
# Check optional fields and include in config if not present
for optkey, value in template_config['optional'].items():
--
GitLab
From b7dc184aac26bf4f017552d82230a6eec5326f4f Mon Sep 17 00:00:00 2001
From: burri0000 <dominik.burri@unibas.ch>
Date: Wed, 14 Jul 2021 11:12:51 +0200
Subject: [PATCH 06/11] separate empty and erroneous optional field.
---
workflow/Snakefile | 9 ++++++---
1 file changed, 6 insertions(+), 3 deletions(-)
diff --git a/workflow/Snakefile b/workflow/Snakefile
index 8ffa423..15faefd 100644
--- a/workflow/Snakefile
+++ b/workflow/Snakefile
@@ -38,14 +38,17 @@ if missing:
logger.error(err_msg)
raise ValueError(err_msg)
# Check if optional field available
-if 'optional' not in config or type(config['optional']) not in [OrderedDict, dict]:
- logger.info(f'No "optional" field found or no valid configuration.')
+if 'optional' not in config:
+ logger.info(f'No "optional" section found, adding dictionary "optional" to configuration.')
config['optional'] = {}
+if type(config['optional']) not in [OrderedDict, dict]:
+ logger.error(f'No valid section "optional" supplied. Got {config["optional"]}, type: {type(config["optional"])} but require dictionary.')
+ raise TypeError
# Check optional fields and include in config if not present
for optkey, value in template_config['optional'].items():
if optkey not in config['optional']:
config['optional'][optkey] = value
- logger.info(f'Set default value for optional parameter "{optkey}" to : "{value}"')
+ logger.info(f'No value for optional parameter "{optkey}" found, set to default: "{value}"')
# Parse YAML rule config file
--
GitLab
From 34622aa1ad5500dec43e71d5daa71ae201d5af64 Mon Sep 17 00:00:00 2001
From: burri0000 <dominik.burri@unibas.ch>
Date: Wed, 14 Jul 2021 13:46:37 +0200
Subject: [PATCH 07/11] properly raise TypeError and rephrase error message.
---
workflow/Snakefile | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/workflow/Snakefile b/workflow/Snakefile
index 15faefd..e55a954 100644
--- a/workflow/Snakefile
+++ b/workflow/Snakefile
@@ -57,7 +57,8 @@ try:
rule_config = yaml.safe_load(_file)
logger.info(f"Loaded rule_config from {config['optional']['rule_config']}.")
except TypeError:
- logger.error(f'No string supplied at field "rule_config", but: {config["optional"]["rule_config"]}')
+ logger.error(f'No string supplied at field "rule_config", but: {type(config["optional"]["rule_config"])} with content: {config["optional"]["rule_config"]}')
+ raise
except FileNotFoundError:
logger.error(f"No rule config file found at {config['optional']['rule_config']}. Either provide file or remove rule_config parameter from config.yaml! ")
raise
--
GitLab
From 0486c79f831996f1ee6c60bdc3d6601b48b6effc Mon Sep 17 00:00:00 2001
From: burri0000 <dominik.burri@unibas.ch>
Date: Wed, 14 Jul 2021 14:53:43 +0200
Subject: [PATCH 08/11] use json schema and snakemakes validation util.
---
resources/config_schema.json | 65 +++++++++++++++++++++
tests/input_files/config.mutliple_lanes.yml | 9 ++-
tests/input_files/config.yaml | 13 ++---
tests/input_files/config_alfa.yaml | 9 ++-
workflow/Snakefile | 55 +++++------------
5 files changed, 92 insertions(+), 59 deletions(-)
create mode 100644 resources/config_schema.json
diff --git a/resources/config_schema.json b/resources/config_schema.json
new file mode 100644
index 0000000..8db7929
--- /dev/null
+++ b/resources/config_schema.json
@@ -0,0 +1,65 @@
+{
+ "$schema": "http://json-schema.org/draft-07/schema#",
+ "title": "Configuration schema",
+ "required": ["samples", "output_dir", "log_dir", "kallisto_indexes", "salmon_indexes", "star_indexes", "alfa_indexes"],
+ "type": "object",
+ "properties": {
+ "samples": {
+ "type": "string",
+ "description": "Path to samples table."
+ },
+ "output_dir": {
+ "type": "string",
+ "description": "Path to output directory."
+ },
+ "log_dir": {
+ "type": "string",
+ "description": "Path to log directory."
+ },
+ "kallisto_indexes": {
+ "type": "string",
+ "description": "Path to kallisto indexes directory."
+ },
+ "salmon_indexes": {
+ "type": "string",
+ "description": "Path to salmon indexes directory."
+ },
+ "star_indexes": {
+ "type": "string",
+ "description": "Path to star indexes directory."
+ },
+ "alfa_indexes": {
+ "type": "string",
+ "description": "Path to alfa indexes directory."
+ },
+ "rule_config": {
+ "type": "string",
+ "description": "Path to rule configuration file."
+ },
+ "report_description": {
+ "type": "string",
+ "description": "Description of the run to appear in the multiqc report.",
+ "default": "No description provided by user."
+ },
+ "report_logo": {
+ "type": "string",
+ "description": "Relative path to image to display as logo in multiqc report.",
+ "default": "../../images/logo.128px.png"
+ },
+ "report_url": {
+ "type": "string",
+ "description": "URL to appear in multiqc report.",
+ "default": "https://zavolan.biozentrum.unibas.ch/"
+ },
+ "author_name": {
+ "type": "string",
+ "description": "Full author name to display in multiqc report.",
+ "default": "NA"
+ },
+ "author_email": {
+ "type": "string",
+ "description": "Author e-mail address to display in multiqc report.",
+ "default": "NA"
+ }
+ }
+}
diff --git a/tests/input_files/config.mutliple_lanes.yml b/tests/input_files/config.mutliple_lanes.yml
index 016692d..8fb3406 100644
--- a/tests/input_files/config.mutliple_lanes.yml
+++ b/tests/input_files/config.mutliple_lanes.yml
@@ -6,9 +6,8 @@
salmon_indexes: "results/salmon_indexes"
star_indexes: "results/star_indexes"
alfa_indexes: "results/alfa_indexes"
- optional:
- rule_config: "../input_files/rule_config.yaml"
- report_description: "No description provided by user"
- report_logo: "../../images/logo.128px.png"
- report_url: "https://zavolan.biozentrum.unibas.ch/"
+ rule_config: "../input_files/rule_config.yaml"
+ report_description: "No description provided by user"
+ report_logo: "../../images/logo.128px.png"
+ report_url: "https://zavolan.biozentrum.unibas.ch/"
...
diff --git a/tests/input_files/config.yaml b/tests/input_files/config.yaml
index 54edf9c..1fc480c 100644
--- a/tests/input_files/config.yaml
+++ b/tests/input_files/config.yaml
@@ -8,11 +8,10 @@
star_indexes: "results/star_indexes"
alfa_indexes: "results/alfa_indexes"
# Optional fields
- optional:
- rule_config: "../input_files/rule_config.yaml"
- report_description: "No description provided by user"
- report_logo: "../../images/logo.128px.png"
- report_url: "https://zavolan.biozentrum.unibas.ch/"
- author_name: "NA"
- author_email: "NA"
+ rule_config: "../input_files/rule_config.yaml"
+ report_description: "No description provided by user"
+ report_logo: "../../images/logo.128px.png"
+ report_url: "https://zavolan.biozentrum.unibas.ch/"
+ author_name: "NA"
+ author_email: "NA"
...
diff --git a/tests/input_files/config_alfa.yaml b/tests/input_files/config_alfa.yaml
index 06decc6..5edcd7e 100644
--- a/tests/input_files/config_alfa.yaml
+++ b/tests/input_files/config_alfa.yaml
@@ -6,9 +6,8 @@
salmon_indexes: "results/salmon_indexes"
star_indexes: "results/star_indexes"
alfa_indexes: "results/alfa_indexes"
- optional:
- rule_config: "../input_files/rule_config.yaml"
- report_description: "No description provided by user"
- report_logo: "../../images/logo.128px.png"
- report_url: "https://zavolan.biozentrum.unibas.ch/"
+ rule_config: "../input_files/rule_config.yaml"
+ report_description: "No description provided by user"
+ report_logo: "../../images/logo.128px.png"
+ report_url: "https://zavolan.biozentrum.unibas.ch/"
...
diff --git a/workflow/Snakefile b/workflow/Snakefile
index e55a954..fc268e4 100644
--- a/workflow/Snakefile
+++ b/workflow/Snakefile
@@ -5,6 +5,7 @@ import shutil
import yaml
from shlex import quote
from typing import Tuple
+from snakemake.utils import validate
## Preparations
# Get sample table
@@ -17,50 +18,20 @@ samples_table = pd.read_csv(
sep="\t",
)
-# Validate config with template config
-template_config_path = os.path.join(
- workflow.current_basedir,
- '..',
- 'tests',
- 'input_files',
- 'config.yaml')
-try:
- with open(template_config_path) as _file:
- template_config = yaml.safe_load(_file)
- logger.info(f"Loaded template config from {template_config_path}.")
-except FileNotFoundError:
- logger.error(f"No config file found at {template_config_path}! ")
- raise
-# Check if config contains required fields in example config
-missing = [key for key in template_config if key not in config and key != 'optional']
-if missing:
- err_msg = f"Required keys missing: {missing}"
- logger.error(err_msg)
- raise ValueError(err_msg)
-# Check if optional field available
-if 'optional' not in config:
- logger.info(f'No "optional" section found, adding dictionary "optional" to configuration.')
- config['optional'] = {}
-if type(config['optional']) not in [OrderedDict, dict]:
- logger.error(f'No valid section "optional" supplied. Got {config["optional"]}, type: {type(config["optional"])} but require dictionary.')
- raise TypeError
-# Check optional fields and include in config if not present
-for optkey, value in template_config['optional'].items():
- if optkey not in config['optional']:
- config['optional'][optkey] = value
- logger.info(f'No value for optional parameter "{optkey}" found, set to default: "{value}"')
-
+# Validat config
+validate(config, os.path.join("..", "resources", "config_schema.json"))
+logger.info(f'Config file after validation: {config}')
# Parse YAML rule config file
try:
- with open(config['optional']['rule_config']) as _file:
+ with open(config['rule_config']) as _file:
rule_config = yaml.safe_load(_file)
- logger.info(f"Loaded rule_config from {config['optional']['rule_config']}.")
+ logger.info(f"Loaded rule_config from {config['rule_config']}.")
except TypeError:
- logger.error(f'No string supplied at field "rule_config", but: {type(config["optional"]["rule_config"])} with content: {config["optional"]["rule_config"]}')
+ logger.error(f'No string supplied at field "rule_config", but: {type(config["rule_config"])} with content: {config["rule_config"]}')
raise
except FileNotFoundError:
- logger.error(f"No rule config file found at {config['optional']['rule_config']}. Either provide file or remove rule_config parameter from config.yaml! ")
+ logger.error(f"No rule config file found at {config['rule_config']}. Either provide file or remove rule_config parameter from config.yaml! ")
raise
except KeyError:
rule_config = {}
@@ -1580,11 +1551,11 @@ rule prepare_multiqc_config:
"multiqc_config.yaml")
params:
- logo_path = config['optional']['report_logo'],
- multiqc_intro_text = config['optional']['report_description'],
- url = config['optional']['report_url'],
- author_name = config['optional']['author_name'],
- author_email = config['optional']['author_email'],
+ logo_path = config['report_logo'],
+ multiqc_intro_text = config['report_description'],
+ url = config['report_url'],
+ author_name = config['author_name'],
+ author_email = config['author_email'],
additional_params = parse_rule_config(
rule_config,
current_rule=current_rule,
--
GitLab
From 4aa9ede05a6148905af6f5ba7f4e372a47754de5 Mon Sep 17 00:00:00 2001
From: burri0000 <dominik.burri@unibas.ch>
Date: Wed, 14 Jul 2021 14:57:38 +0200
Subject: [PATCH 09/11] cleanup remaining optional keywords
---
scripts/prepare_inputs.py | 7 +++----
1 file changed, 3 insertions(+), 4 deletions(-)
diff --git a/scripts/prepare_inputs.py b/scripts/prepare_inputs.py
index e47fa3d..426902a 100755
--- a/scripts/prepare_inputs.py
+++ b/scripts/prepare_inputs.py
@@ -678,10 +678,9 @@ def main(args):
salmon_indexes: "{salmon_indexes}"
star_indexes: "{star_indexes}"
alfa_indexes: "{alfa_indexes}"
- optional:
- report_description: "{args.description}"
- report_logo: "{args.logo}"
- report_url: "{args.url}"
+ report_description: "{args.description}"
+ report_logo: "{args.logo}"
+ report_url: "{args.url}"
...
'''
args.config_file.write(config_file_content)
--
GitLab
From 39baeaa417f4aa032b848aff248936b575b99818 Mon Sep 17 00:00:00 2001
From: burri0000 <dominik.burri@unibas.ch>
Date: Wed, 14 Jul 2021 15:10:11 +0200
Subject: [PATCH 10/11] update md5sum
---
tests/test_scripts_prepare_inputs_table/expected_output.md5 | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/tests/test_scripts_prepare_inputs_table/expected_output.md5 b/tests/test_scripts_prepare_inputs_table/expected_output.md5
index dc555cc..53afed5 100644
--- a/tests/test_scripts_prepare_inputs_table/expected_output.md5
+++ b/tests/test_scripts_prepare_inputs_table/expected_output.md5
@@ -1,2 +1,2 @@
-90e90bb60335a2e17cd40cbf058af7b7 config.yaml
+40bd0f0fcecdd0d9bc932f63c2811478 config.yaml
c8dcc5a203e9046806c4090525960151 samples.tsv
--
GitLab
From b0c6af80e49ac2c88e3c6e410697544ee2b44b0e Mon Sep 17 00:00:00 2001
From: burri0000 <dominik.burri@unibas.ch>
Date: Thu, 15 Jul 2021 14:56:38 +0200
Subject: [PATCH 11/11] do not allow additional properties in config
---
resources/config_schema.json | 1 +
1 file changed, 1 insertion(+)
diff --git a/resources/config_schema.json b/resources/config_schema.json
index 8db7929..1ad7846 100644
--- a/resources/config_schema.json
+++ b/resources/config_schema.json
@@ -3,6 +3,7 @@
"title": "Configuration schema",
"required": ["samples", "output_dir", "log_dir", "kallisto_indexes", "salmon_indexes", "star_indexes", "alfa_indexes"],
"type": "object",
+ "additionalProperties": false,
"properties": {
"samples": {
"type": "string",
--
GitLab