Merge pull request #17 from NBISweden/plugin

MultiQC plugin

README.md

@@ -91,7 +91,21 @@ Install blr into the environment in "editable install" mode.
 This will install blr in such a way that you can still modify the source code
 and get any changes immediately without re-installing.
 
-#### 2.1 Linux users (not macOS)
+#### 2.1 MultiQC plugin
+
+There is a MultiQC plugin included in the BLR pipeline called 
+MultiQC_BLR. If you which to run MultiQC without this plugin include 
+`--disable-blr-plugin` in your multiqc command. 
+
+The plugin allows for comparision between different runs. In this case go to 
+the directory containing the folders for the runs you wish to compare. Then run:
+
+    multiqc -d .
+
+The `-d` option prepends the directory name to each sample allowing differentiation 
+between the runs. 
+
+#### 2.2 Linux users (not macOS)
 
 To enable DeepVariant, install it separately to your environment.
 

setup.py

@@ -18,14 +18,29 @@
         "tqdm",
         "snakemake",
         "importlib_resources; python_version<'3.7'",
+        "multiqc"
     ],
     extras_require={
         "dev": ["flake8"],
     },
     package_dir={"": "src"},
     packages=find_namespace_packages("src"),
     package_data={"blr": ["Snakefile", "rules/*.smk", "config.schema.yaml", "blr.yaml"]},
-    entry_points={"console_scripts": ["blr = blr.__main__:main"]},
+    entry_points={
+        "console_scripts": [
+            "blr = blr.__main__:main"
+        ],
+        "multiqc.modules.v1": [
+            "stats = multiqc_blr.modules.stats:MultiqcModule",
+            "hapcut2 = multiqc_blr.modules.hapcut2:MultiqcModule",
+        ],
+        "multiqc.cli_options.v1": [
+            "disable_plugin = multiqc_blr.cli:disable_plugin"
+        ],
+        "multiqc.hooks.v1": [
+            "execution_start = multiqc_blr.multiqc_blr:execution_start"
+        ]
+    },
     classifiers=[
         "Development Status :: 4 - Beta",
         "License :: OSI Approved :: MIT License",

src/blr/Snakefile

@@ -66,7 +66,7 @@ def get_multiqc_input(wildcards):
         "figures/mapped",
     ]
     if config["phasing_ground_truth"]:
-        inputs.append("mapped.phasing_stats.tsv")
+        inputs.append("mapped.phasing_stats.txt")
 
     return inputs
 

src/blr/cli/calculate_haplotype_statistics.py

@@ -21,12 +21,11 @@ def main(args):
     logging.info("Starting analysis")
     stats = vcf_vcf_error_rate(args.vcf1, args.vcf2, args.indels)
 
-    with open(args.output + ".txt", "w") as file:
-        print(stats.to_txt(), file=file)
-
-    with open(args.output + ".tsv", "w") as file:
-        print(stats.to_tsv(), file=file)
-
+    if args.output:
+        with open(args.output, "w") as file:
+            print(stats.to_txt(), file=file)
+    else:
+        print(stats.to_txt())
     logging.info("Finished")
 
 
@@ -639,4 +638,4 @@ def add_arguments(parser):
                         help="A phased, single sample  VCF file to use as the 'ground truth' haplotype.")
     parser.add_argument('-i', '--indels', action="store_true",
                         help='Use this flag to consider indel variants. Default: %(default)s', default=False)
-    parser.add_argument("-o", "--output", default="phasing_stats", help="Output file prefix. Default: %(default)s.")
+    parser.add_argument("-o", "--output", help="Output file name. Default: Print to stdout.")

src/blr/multiqc_config.yaml

@@ -1,56 +1,3 @@
 title: "BLR run report"
 intro_text: "This is a collection of results from the BLR pipeline. Go to our
 <a href=https://github.com/FrickTobias/BLR>github page</a> for more information."
-
-custom_data:
-    phasing_stats:
-        id: 'phasing_stats'
-        section_name: 'Phasing metrics'
-        description: '- Phasing statistics from HapCUP2 calculate_phasing_stats.py script.'
-        plot_type: 'table'
-        pconfig:
-            id: 'phasing_stats'
-            title: 'Phasing metrics table'
-            scale: false
-            share_key: false
-        headers:
-            'switch rate':
-                title: 'Switch rate'
-                description: 'switch errors as a fraction of possible positions for switch errors'
-                format: '{:,.7f}'
-                placement: 1
-            'mismatch rate':
-                title: 'Mismatch rate'
-                description: 'mismatch errors as a fraction of possible positions for mismatch errors'
-                format: '{:,.7f}'
-                placement: 2
-            'flat rate':
-                title: 'Flat rate'
-                description: 'flat errors as a fraction of possible positions for flat errors'
-                format: '{:,.7f}'
-                hidden: true
-            'phased count':
-                title: 'Phased count'
-                description: 'count of total SNVs phased in the test haplotype'
-                format: '{:,.0f}'
-                placement: 3
-            AN50:
-                title: 'AN50 (Mbp)'
-                description: 'the AN50 metric of haplotype completeness'
-                format: '{:,.3f}'
-                hidden: true
-            N50:
-                title: 'N50 (Mbp)'
-                description: 'the N50 metric of haplotype completeness'
-                format: '{:,.3f}'
-                placement: 4
-            'num snps max blk':
-                title: 'SNPs in max blk'
-                description: 'the fraction of SNVs in the largest (most variants phased) block'
-                format: '{:,.0f}'
-                placement: 5
-
-
-sp:
-    phasing_stats:
-        fn: 'phasing_stats.tsv'

src/blr/rules/phasing.smk

@@ -68,17 +68,15 @@ rule hapcut2_stats:
     """Get phasing statistics relative the ground truth. See https://github.com/vibansal/HapCUT2/tree/master/utilities
     for details. """
     output:
-        stats = expand("{{base}}.phasing_stats.{ext}", ext=["txt", "tsv"])
+        stats = "{base}.phasing_stats.txt"
     input:
         vcf1 = "{base}.calling.phased.vcf",
         vcf2 = "ground_truth.phased.vcf"
-    params:
-        stats_prefix = "{base}.phasing_stats"
     shell:
         "blr calculate_haplotype_statistics"
         " -v1 {input.vcf1}"
         " -v2 {input.vcf2}"
-        " -o {params.stats_prefix}"
+        " -o {output.stats}"
 
 
 def get_haplotag_input(wildcards):

src/multiqc_blr/cli.py

@@ -0,0 +1,16 @@
+#!/usr/bin/env python
+"""
+MultiQC command line options - we tie into the MultiQC
+core here and add some new command line parameters.
+
+See the Click documentation for more command line flag types:
+http://click.pocoo.org/5/
+"""
+
+import click
+
+# Sets config.kwargs['disable_plugin'] to True if specified (will be False otherwise)
+disable_plugin = click.option(
+    '--disable-blr-plugin', 'disable_plugin',
+    is_flag=True,
+    help="Disable the MultiQC BLR plugin on this run")

src/multiqc_blr/modules/hapcut2/__init__.py

@@ -0,0 +1,3 @@
+from __future__ import absolute_import
+
+from .hapcut2 import MultiqcModule  # noqa