revisit demultiplexing, sv cli commands

harpy/_conda.py

@@ -32,6 +32,11 @@ def create_conda_recipes(outdir: str, envs: list=None) -> None:
             "bioconda::samtools",
             "bioconda::tigmint"
         ],
+        "demultiplex": [
+            "bioconda::pheniqs",
+            "bioconda::pysam",
+            "conda-forge::python-levenshtein"
+        ],
         "metassembly": [
             "bioconda::athena_meta=1.2"
         ],

harpy/_printing.py

@@ -100,3 +100,16 @@ def print_onerror(logfile):
     console.rule("[bold]Workflow Error", style = "red")
     console.print(f"The workflow stopped because of an error. Full workflow log:\n[bold]{logfile}[/bold]")
     console.rule("[bold]Where Error Occurred", style = "red")
+
+def workflow_info(*arg):
+    """
+    Accepts an unlimited number of length-2 lists or tuples and returns a rich.Table with the value of the first indices as the row names and the second indices as the values
+    Use None instead of a list to ignore that entry (useful for conditionals). The second value will always be converted to a string.
+    """
+    table = Table(show_header=False,pad_edge=False, show_edge=False, padding = (0,0), box=box.SIMPLE)
+    table.add_column("detail", justify="left", style="light_steel_blue", no_wrap=True)
+    table.add_column("value", justify="left")
+    for i in arg:
+        if arg:
+            table.add_row(i[0], str(i[1]))
+    return table

harpy/align.py

@@ -4,16 +4,14 @@
 import sys
 import yaml
 from pathlib import Path
-from rich import box
-from rich.table import Table
 import rich_click as click
 from ._conda import create_conda_recipes
 from ._misc import fetch_report, fetch_rule, snakemake_log
 from ._cli_types_generic import  ContigList, InputFile, HPCProfile, SnakemakeParams
 from ._cli_types_params import BwaParams, EmaParams, StrobeAlignParams
 from ._launch import launch_snakemake, SNAKEMAKE_CMD
 from ._parsers import parse_fastq_inputs
-from ._printing import print_error, print_solution, print_notice
+from ._printing import print_error, print_solution, print_notice, workflow_info
 from ._validations import check_fasta, fasta_contig_match, validate_barcodefile
 
 @click.group(options_metavar='', context_settings={"help_option_names" : ["-h", "--help"]})
@@ -141,13 +139,12 @@ def bwa(inputs, output_dir, genome, depth_window, threads, keep_unmapped, extra_
     if setup_only:
         sys.exit(0)
 
-    start_text = Table(show_header=False,pad_edge=False, show_edge=False, padding = (0,0), box=box.SIMPLE)
-    start_text.add_column("detail", justify="left", style="light_steel_blue", no_wrap=True)
-    start_text.add_column("value", justify="left")
-    start_text.add_row("Samples:", f"{sample_count}")
-    start_text.add_row("Genome:", genome)
-    start_text.add_row("Output Folder:", output_dir + "/")
-    start_text.add_row("Workflow Log:", sm_log.replace(f"{output_dir}/", "") + "[dim].gz")
+    start_text = (
+        ("Samples:",sample_count),
+        ("Genome:", genome),
+        ("Output Folder:", output_dir + "/"),
+        ("Workflow Log:", sm_log.replace(f"{output_dir}/", "") + "[dim].gz")
+    )
     launch_snakemake(command, "align_bwa", start_text, output_dir, sm_log, quiet, "workflow/align.bwa.summary")
 
 @click.command(no_args_is_help = True, context_settings=dict(allow_interspersed_args=False), epilog = "Documentation: https://pdimens.github.io/harpy/workflows/align/ema")
@@ -251,14 +248,13 @@ def ema(inputs, output_dir, platform, barcode_list, fragment_density, genome, de
     if setup_only:
         sys.exit(0)
 
-    start_text = Table(show_header=False,pad_edge=False, show_edge=False, padding = (0,0), box=box.SIMPLE)
-    start_text.add_column("detail", justify="left", style="light_steel_blue", no_wrap=True)
-    start_text.add_column("value", justify="left")
-    start_text.add_row("Samples:", f"{sample_count}")
-    start_text.add_row("Genome:", genome)
-    start_text.add_row("Platform:", platform)
-    start_text.add_row("Output Folder:", output_dir + "/")
-    start_text.add_row("Workflow Log:", sm_log.replace(f"{output_dir}/", "") + "[dim].gz")
+    start_text = workflow_info(
+        ("Samples:",sample_count),
+        ("Genome:", genome),
+        ("Platform:", platform),
+        ("Output Folder:", output_dir + "/"),
+        ("Workflow Log:", sm_log.replace(f"{output_dir}/", "") + "[dim].gz")
+    )
     launch_snakemake(command, "align_ema", start_text, output_dir, sm_log, quiet, "workflow/align.ema.summary")
 
 @click.command(no_args_is_help = True, epilog= "Documentation: https://pdimens.github.io/harpy/workflows/align/strobe/")
@@ -343,13 +339,12 @@ def strobe(inputs, output_dir, genome, read_length, keep_unmapped, depth_window,
     if setup_only:
         sys.exit(0)
 
-    start_text = Table(show_header=False,pad_edge=False, show_edge=False, padding = (0,0), box=box.SIMPLE)
-    start_text.add_column("detail", justify="left", style="light_steel_blue", no_wrap=True)
-    start_text.add_column("value", justify="left")
-    start_text.add_row("Samples:", f"{sample_count}")
-    start_text.add_row("Genome:", genome)
-    start_text.add_row("Output Folder:", output_dir + "/")
-    start_text.add_row("Workflow Log:", sm_log.replace(f"{output_dir}/", "") + "[dim].gz")
+    start_text = workflow_info(
+        ("Samples:",sample_count),
+        ("Genome:", genome),
+        ("Output Folder:", output_dir + "/"),
+        ("Workflow Log:", sm_log.replace(f"{output_dir}/", "") + "[dim].gz")
+    )
     launch_snakemake(command, "align_strobe", start_text, output_dir, sm_log, quiet, "workflow/align.strobealign.summary")
 
 align.add_command(bwa)

harpy/assembly.py

@@ -3,14 +3,13 @@
 import os
 import sys
 import yaml
-from rich import box
-from rich.table import Table
 import rich_click as click
 from ._cli_types_generic import KParam, HPCProfile, SnakemakeParams
 from ._cli_types_params import SpadesParams, ArcsParams
 from ._conda import create_conda_recipes
 from ._launch import launch_snakemake, SNAKEMAKE_CMD
 from ._misc import fetch_rule, snakemake_log
+from ._printing import workflow_info
 from ._validations import validate_fastq_bx
 
 docstring = {
@@ -129,14 +128,10 @@ def assembly(fastq_r1, fastq_r2, bx_tag, kmer_length, max_memory, output_dir, ex
     if setup_only:
         sys.exit(0)
 
-    start_text = Table(show_header=False,pad_edge=False, show_edge=False, padding = (0,0), box=box.SIMPLE)
-    start_text.add_column("detail", justify="left", style="light_steel_blue", no_wrap=True)
-    start_text.add_column("value", justify="left")
-    start_text.add_row("Barcode Tag: ", bx_tag.upper())
-    if kmer_length == "auto":
-        start_text.add_row("Kmer Length: ", "auto")
-    else:
-        start_text.add_row("Kmer Length: ", ",".join(map(str,kmer_length)))
-    start_text.add_row("Output Folder:", f"{output_dir}/")
-    start_text.add_row("Workflow Log:", sm_log.replace(f"{output_dir}/", "") + "[dim].gz")
+    start_text = workflow_info(
+        ("Barcode Tag: ", bx_tag.upper()),
+        ("Kmer Length: ", "auto") if kmer_length == "auto" else ("Kmer Length: ", ",".join(map(str,kmer_length))),
+        ("Output Folder:", f"{output_dir}/"),
+        ("Workflow Log:", sm_log.replace(f"{output_dir}/", "") + "[dim].gz")
+    )
     launch_snakemake(command, asm, start_text, output_dir, sm_log, quiet, f"workflow/{asm}.summary")

harpy/bin/haplotag_acbd.py

@@ -27,7 +27,7 @@
 }
 
 for BC in ["A","C","B","D"]:
-    with open(f"{outdir}/BC_{BC}.txt", "w", encoding="utf-8") as f:
+    with open(f"{outdir}/segment_{BC}.bc", "w", encoding="utf-8") as f:
         ID = [f"{BC}{number:02d}" for number in range(1, 97)]
         delim = ["	".join(tup) for tup in zip(ID, BX[BC])]
         _ = [f.write(f"{i}\n") for i in delim]

harpy/deconvolve.py

@@ -3,14 +3,13 @@
 import os
 import sys
 import yaml
-from rich import box
-from rich.table import Table
 import rich_click as click
 from ._cli_types_generic import HPCProfile, SnakemakeParams
 from ._conda import create_conda_recipes
 from ._launch import launch_snakemake, SNAKEMAKE_CMD
 from ._misc import fetch_rule, snakemake_log
 from ._parsers import parse_fastq_inputs
+from ._printing import workflow_info
 
 docstring = {
     "harpy deconvolve": [
@@ -84,10 +83,9 @@ def deconvolve(inputs, output_dir, kmer_length, window_size, density, dropout, t
     if setup_only:
         sys.exit(0)
 
-    start_text = Table(show_header=False,pad_edge=False, show_edge=False, padding = (0,0), box=box.SIMPLE)
-    start_text.add_column("detail", justify="left", style="light_steel_blue", no_wrap=True)
-    start_text.add_column("value", justify="left")
-    start_text.add_row("Samples:", f"{sample_count}")
-    start_text.add_row("Output Folder:", output_dir + "/")
-    start_text.add_row("Workflow Log:", sm_log.replace(f"{output_dir}/", "") + "[dim].gz")
+    start_text = workflow_info(
+        ("Samples:", sample_count),
+        ("Output Folder:", output_dir + "/"),
+        ("Workflow Log:", sm_log.replace(f"{output_dir}/", "") + "[dim].gz")
+    )
     launch_snakemake(command, "deconvolve", start_text, output_dir, sm_log, quiet, "workflow/deconvolve.summary")