move ignore-bx to workflow controls, add it as an option to qc

Author: pdimens

harpy/align.py

@@ -33,11 +33,11 @@ def align():
     "harpy align bwa": [
         {
             "name": "Parameters",
-            "options": ["--extra-params", "--genome", "--ignore-bx", "--keep-unmapped", "--molecule-distance", "--min-quality"],
+            "options": ["--extra-params", "--genome", "--keep-unmapped", "--molecule-distance", "--min-quality"],
         },
         {
             "name": "Workflow Controls",
-            "options": ["--container", "--contigs", "--depth-window", "--hpc", "--output-dir", "--quiet", "--skip-reports", "--snakemake", "--threads", "--help"],
+            "options": ["--container", "--contigs", "--depth-window", "--hpc", "--ignore-bx", "--output-dir", "--quiet", "--skip-reports", "--snakemake", "--threads", "--help"],
         },
     ],
     "harpy align ema": [
@@ -53,11 +53,11 @@ def align():
     "harpy align strobe": [
         {
             "name": "Parameters",
-            "options": ["--extra-params", "--genome", "--ignore-bx", "--keep-unmapped", "--molecule-distance", "--min-quality", "--read-length"],
+            "options": ["--extra-params", "--genome", "--keep-unmapped", "--molecule-distance", "--min-quality", "--read-length"],
         },
         {
             "name": "Workflow Controls",
-            "options": ["--container", "--contigs", "--depth-window", "--hpc", "--output-dir", "--quiet", "--skip-reports", "--snakemake", "--threads", "--help"],
+            "options": ["--container", "--contigs", "--depth-window", "--hpc", "--ignore-bx", "--output-dir", "--quiet", "--skip-reports", "--snakemake", "--threads", "--help"],
         },
     ]
 }
@@ -66,7 +66,6 @@ def align():
 @click.option('-g', '--genome', type=InputFile("fasta", gzip_ok = True), required = True, help = 'Genome assembly for read mapping')
 @click.option('-w', '--depth-window', default = 50000, show_default = True, type = int, help = 'Interval size (in bp) for depth stats')
 @click.option('-x', '--extra-params', type = BwaParams(), help = 'Additional bwa mem parameters, in quotes')
-@click.option('-i', '--ignore-bx',  is_flag = True, default = False, help = 'Ignore parts of the workflow specific to linked-read sequences')
 @click.option('-u', '--keep-unmapped',  is_flag = True, default = False, help = 'Retain unmapped sequences in the output')
 @click.option('-q', '--min-quality', default = 30, show_default = True, type = click.IntRange(min = 0, max = 40), help = 'Minimum mapping quality to pass filtering')
 @click.option('-d', '--molecule-distance', default = 100000, show_default = True, type = click.IntRange(min = 0, max_open = True), help = 'Distance cutoff to split molecules (bp)')
@@ -76,6 +75,7 @@ def align():
 @click.option('--contigs',  type = ContigList(), help = 'File or list of contigs to plot')
 @click.option('--setup-only',  is_flag = True, hidden = True, default = False, help = 'Setup the workflow and exit')
 @click.option('--hpc',  type = HPCProfile(), help = 'Directory with HPC submission `config.yaml` file')
+@click.option('--ignore-bx',  is_flag = True, default = False, help = 'Ignore parts of the workflow specific to linked-read sequences')
 @click.option('--quiet',  is_flag = True, show_default = True, default = False, help = 'Don\'t show output text while running')
 @click.option('--skip-reports',  is_flag = True, show_default = True, default = False, help = 'Don\'t generate HTML reports')
 @click.option('--snakemake', type = SnakemakeParams(), help = 'Additional Snakemake parameters, in quotes')
@@ -264,7 +264,6 @@ def ema(inputs, output_dir, platform, barcode_list, fragment_density, genome, de
 @click.option('-g', '--genome', type=InputFile("fasta", gzip_ok = True), required = True, help = 'Genome assembly for read mapping')
 @click.option('-w', '--depth-window', default = 50000, show_default = True, type = int, help = 'Interval size (in bp) for depth stats')
 @click.option('-x', '--extra-params', type = StrobeAlignParams(), help = 'Additional aligner parameters, in quotes')
-@click.option('-i', '--ignore-bx',  is_flag = True, default = False, help = 'Ignore parts of the workflow specific to linked-read sequences')
 @click.option('-u', '--keep-unmapped',  is_flag = True, default = False, help = 'Retain unmapped sequences in the output')
 @click.option('-q', '--min-quality', default = 30, show_default = True, type = click.IntRange(min = 0, max = 40), help = 'Minimum mapping quality to pass filtering')
 @click.option('-d', '--molecule-distance', default = 100000, show_default = True, type = click.IntRange(min = 0, max_open = True), help = 'Distance cutoff to split molecules (bp)')
@@ -275,6 +274,7 @@ def ema(inputs, output_dir, platform, barcode_list, fragment_density, genome, de
 @click.option('--container',  is_flag = True, default = False, help = 'Use a container instead of conda')
 @click.option('--setup-only',  is_flag = True, hidden = True, default = False, help = 'Setup the workflow and exit')
 @click.option('--hpc',  type = HPCProfile(), help = 'Directory with HPC submission `config.yaml` file')
+@click.option('--ignore-bx',  is_flag = True, default = False, help = 'Ignore parts of the workflow specific to linked-read sequences')
 @click.option('--quiet',  is_flag = True, show_default = True, default = False, help = 'Don\'t show output text while running')
 @click.option('--skip-reports',  is_flag = True, show_default = True, default = False, help = 'Don\'t generate HTML reports')
 @click.option('--snakemake', type = SnakemakeParams(), help = 'Additional Snakemake parameters, in quotes')

harpy/qc.py

@@ -21,7 +21,7 @@
         },
         {
             "name": "Workflow Controls",
-            "options": ["--container", "--hpc", "--output-dir", "--quiet", "--skip-reports", "--snakemake", "--threads", "--help"],
+            "options": ["--container", "--hpc", "--ignore-bx", "--output-dir", "--quiet", "--skip-reports", "--snakemake", "--threads", "--help"],
         },
     ]
 }
@@ -38,11 +38,12 @@
 @click.option('--container',  is_flag = True, default = False, help = 'Use a container instead of conda')
 @click.option('--setup-only',  is_flag = True, hidden = True, show_default = True, default = False, help = 'Setup the workflow and exit')
 @click.option('--hpc',  type = HPCProfile(), help = 'Directory with HPC submission `config.yaml` file')
+@click.option('--ignore-bx',  is_flag = True, default = False, help = 'Ignore parts of the workflow specific to linked-read sequences')
 @click.option('--quiet',  is_flag = True, default = False, help = 'Don\'t show output text while running')
 @click.option('--skip-reports',  is_flag = True, default = False, help = 'Don\'t generate HTML reports')
 @click.option('--snakemake', type = SnakemakeParams(), help = 'Additional Snakemake parameters, in quotes')
 @click.argument('inputs', required=True, type=click.Path(exists=True, readable=True), nargs=-1)
-def qc(inputs, output_dir, min_length, max_length, trim_adapters, deduplicate, deconvolve, extra_params, threads, snakemake, skip_reports, quiet, hpc, container, setup_only):
+def qc(inputs, output_dir, min_length, max_length, trim_adapters, deduplicate, deconvolve, extra_params, ignore_bx, threads, snakemake, skip_reports, quiet, hpc, container, setup_only):
     """
     Remove adapters and quality-control sequences
 
@@ -92,6 +93,7 @@ def qc(inputs, output_dir, min_length, max_length, trim_adapters, deduplicate, d
         "workflow" : "qc",
         "snakemake_log" : sm_log,
         "output_directory" : output_dir,
+        "ignore_bx" : ignore_bx,
         "trim_adapters" : trim_adapters,
         "deduplicate" : deduplicate,
         "min_len" : min_length,

harpy/snakefiles/qc.smk

@@ -19,6 +19,7 @@ envdir       = os.path.join(os.getcwd(), outdir, "workflow", "envs")
 min_len 	 = config["min_len"]
 max_len 	 = config["max_len"]
 extra 	     = config.get("extra", "") 
+ignore_bx    = config["ignore_bx"]
 trim_adapters = config.get("trim_adapters", None)
 dedup        = config["deduplicate"]
 deconvolve   = config.get("deconvolve", False)
@@ -195,7 +196,7 @@ rule workflow_summary:
     default_target: True
     input:
         fq = collect(outdir + "/{sample}.{FR}.fq.gz", FR = ["R1", "R2"], sample = samplenames),
-        bx_report = outdir + "/reports/barcode.summary.html" if not skip_reports else [],
+        bx_report = outdir + "/reports/barcode.summary.html" if not skip_reports and not ignore_bx else [],
         agg_report = outdir + "/reports/qc.report.html" if not skip_reports else []    
     params:
         minlen = f"--length_required {min_len}",