fix the code for new entry point

.github/workflows/createrelease.yml

@@ -35,7 +35,7 @@ jobs:
         run: rm -rf /opt/hostedtoolcache
       - name: Recreate container
         shell: micromamba-shell {0}
-        run: snakemake -s resources/container/containerize.smk --containerize > Dockerfile
+        run: harpy containerize
       - name: Set up Docker Buildx
         uses: docker/setup-buildx-action@v3
       - name: Login to Docker Hub

.github/workflows/rebuildcontainer.yml

@@ -53,11 +53,17 @@ jobs:
           environment-file: resources/harpy.yaml
           cache-environment: true
           post-cleanup: 'all'
+      - name: Install harpy
+        shell: micromamba-shell {0}
+        run: |
+          python3 -m pip install --upgrade build && python3 -m build
+          pip install dist/*.whl
+          resources/buildforCI.sh
       - name: Clear space
         run: rm -rf /opt/hostedtoolcache
       - name: Rebuild dockerfile
         shell: micromamba-shell {0}
-        run: snakemake -s src/harpy/containerize.smk --containerize > Dockerfile
+        run: harpy containerize
       - name: Set up Docker Buildx
         uses: docker/setup-buildx-action@v3
       - name: Login to Docker Hub

src/harpy/__main__.py

@@ -9,12 +9,11 @@
     enverror = "\033[1;33mERROR:\033[00m Harpy expects to run from within an active conda environment, but one was not detected."
     print(enverror, file = sys.stderr)
     fix = "\033[1;34mSOLUTION:\033[00m Activate the conda environment Harpy was installed into and run Harpy again."
-    print()
-    print(fix, file = sys.stderr)
+    print("\n" + fix, file = sys.stderr)
     print(f"\n\033[1mDetails:\033[00m")
     details = "In order to work correctly, Harpy expects several software packages to be available in the PATH, which are provided automatically with Harpy's conda-based installation. It also expects snakefiles, scripts, utilities, etc. to be in the /bin/ folder within that conda environment. If you're seeing this message, no active conda environment was detected upon execution, and Harpy exited as an early failsafe against unexpected runtime errors associated with \"missing\" files and packages."
     print(details, file = sys.stderr)
-    exit(1)
+    sys.exit(1)
 
 from . import align
 from . import demultiplex
@@ -25,11 +24,10 @@
 from . import simulate
 from . import snp
 from . import sv
+from . import container
 from .popgroups import popgroup
 from .stitchparams import stitchparams
-from .conda_deps import generate_conda_deps
 import rich_click as click
-import subprocess
 
 click.rich_click.USE_MARKDOWN = True
 click.rich_click.SHOW_ARGUMENTS = False
@@ -54,7 +52,6 @@ def cli():
     
     **Documentation**: [https://pdimens.github.io/harpy/](https://pdimens.github.io/harpy/)
     """
-    pass
 
 # main program
 #cli.add_command(hpc)
@@ -69,6 +66,7 @@ def cli():
 cli.add_command(impute.impute)
 cli.add_command(phase.phase)
 cli.add_command(simulate.simulate)
+cli.add_command(container.containerize)
 # demultiplex submodules
 demultiplex.demultiplex.add_command(demultiplex.gen1)
 # preflight submodules
@@ -119,18 +117,3 @@ def cli():
 }
 
 click.rich_click.OPTION_GROUPS = demultiplex.docstring | preflight.docstring | qc.docstring | align.docstring | snp.docstring | sv.docstring | impute.docstring | phase.docstring | simulate.docstring
-
-def main():
-    try:
-        workflow = cli(standalone_mode = False)
-        if workflow == 0:
-            return 0
-        elif workflow is not None:
-            generate_conda_deps()
-            _module = subprocess.run(workflow, check = False)
-            return _module.returncode
-    except:
-        sys.exit(1)
-
-if __name__ == '__main__':
-    sys.exit(main())

src/harpy/align.py

@@ -2,8 +2,10 @@
 
 import os
 import sys
+import subprocess
 from time import sleep
 import rich_click as click
+from .conda_deps import generate_conda_deps
 from .helperfunctions import fetch_report, fetch_rule, fetch_script
 from .fileparsers import get_samples_from_fastq, parse_fastq_inputs
 from .printfunctions import print_error, print_solution, print_notice, print_onstart
@@ -114,7 +116,7 @@ def bwa(inputs, output_dir, genome, depth_window, threads, extra_params, quality
     fetch_script(workflowdir, "bxStats.py")
     fetch_report(workflowdir, "AlignStats.Rmd")
 
-    with open(f"{workflowdir}/config.yml", "w", encoding="uft-8") as config:
+    with open(f"{workflowdir}/config.yml", "w", encoding="utf-8") as config:
         config.write(f"genomefile: {genome}\n")
         config.write(f"seq_directory: {workflowdir}/input\n")
         config.write(f"output_directory: {output_dir}\n")
@@ -131,7 +133,9 @@ def bwa(inputs, output_dir, genome, depth_window, threads, extra_params, quality
         f"Samples: {len(samplenames)}\nOutput Directory: {output_dir}",
         "align bwa"
     )
-    return command
+    generate_conda_deps()
+    _module = subprocess.run(command)
+    sys.exit(_module.returncode)
 
 @click.command(no_args_is_help = True, epilog = "read the docs for more information: https://pdimens.github.io/harpy/modules/align/ema")
 @click.option('-p', '--platform', type = click.Choice(['haplotag', '10x'], case_sensitive=False), default = "haplotag", show_default=True, help = "Linked read bead technology\n[haplotag, 10x]")
@@ -223,7 +227,9 @@ def ema(inputs, output_dir, platform, whitelist, genome, depth_window, threads,
         f"Samples: {len(samplenames)}\nPlatform: {platform}\nOutput Directory: {output_dir}/",
         "align ema"
     )
-    return command
+    generate_conda_deps()
+    _module = subprocess.run(command)
+    sys.exit(_module.returncode)
 
 @click.command(no_args_is_help = True, epilog= "read the docs for more information: https://pdimens.github.io/harpy/modules/align/minimap/")
 @click.option('-g', '--genome', type=click.Path(exists=True, dir_okay=False), required = True, help = 'Genome assembly for read mapping')
@@ -294,4 +300,6 @@ def minimap(inputs, output_dir, genome, depth_window, threads, extra_params, qua
         f"Samples: {len(samplenames)}\nOutput Directory: {output_dir}",
         "align minimap"
     )
-    return command
+    generate_conda_deps()
+    _module = subprocess.run(command)
+    sys.exit(_module.returncode)

src/harpy/container.py

@@ -0,0 +1,25 @@
+"""Command to regenerate Dockerfile for container building"""
+
+import os
+import sys
+import subprocess
+import rich_click as click
+from .conda_deps import generate_conda_deps
+from .helperfunctions import fetch_rule
+
+@click.command(no_args_is_help = False, hidden = True)
+def containerize():
+    """
+    Configure conda and docker environments
+
+    **INTERNAL USE ONLY**. Used to recreate all the conda environments required
+    by the workflows and build a dockerfile from that.
+    """
+    generate_conda_deps()
+    fetch_rule(os.getcwd(), "containerize.smk")
+    command = 'snakemake -s containerize.smk --containerize'
+
+    with open("Dockerfile", "w", encoding = "utf-8") as dockerfile:
+        _module = subprocess.run(command.split(), stdout = dockerfile)
+    os.remove("containerize.smk")
+    sys.exit(_module.returncode)

src/harpy/demultiplex.py

@@ -1,7 +1,9 @@
 """Harpy demultiplex workflows"""
 
 import sys
+import subprocess
 import rich_click as click
+from .conda_deps import generate_conda_deps
 from .printfunctions import print_onstart
 from .helperfunctions import fetch_rule
 from .validations import validate_demuxschema
@@ -91,4 +93,6 @@ def gen1(r1_fq, r2_fq, i1_fq, i2_fq, output_dir, schema, threads, snakemake, ski
         f"Haplotag Type: Generation I\nDemultiplex Schema: {schema}\nOutput Directory: {output_dir}",
         "demultiplex gen1"
     )
-    return command
+    generate_conda_deps()
+    _module = subprocess.run(command)
+    sys.exit(_module.returncode)

src/harpy/impute.py

@@ -2,7 +2,9 @@
 
 import os
 import sys
+import subprocess
 import rich_click as click
+from .conda_deps import generate_conda_deps
 from .helperfunctions import fetch_rule, fetch_report, fetch_script, biallelic_contigs
 from .fileparsers import parse_alignment_inputs
 from .printfunctions import print_onstart
@@ -99,4 +101,6 @@ def impute(inputs, output_dir, parameters, threads, vcf, vcf_samples, extra_para
         f"Input VCF: {vcf}\nSamples in VCF: {len(samplenames)}\nAlignments Provided: {len(sn)}\nContigs Considered: {len(contigs)}\nOutput Directory: {output_dir}/",
         "impute"
     )
-    return command
+    generate_conda_deps()
+    _module = subprocess.run(command)
+    sys.exit(_module.returncode)

src/harpy/phase.py

@@ -4,6 +4,7 @@
 import sys
 import subprocess
 import rich_click as click
+from .conda_deps import generate_conda_deps
 from .helperfunctions import fetch_rule, fetch_report
 from .fileparsers import parse_alignment_inputs
 from .printfunctions import print_onstart
@@ -100,4 +101,6 @@ def phase(inputs, output_dir, vcf, threads, molecule_distance, prune_threshold,
         f"Input VCF: {vcf}\nSamples in VCF: {len(samplenames)}\nAlignments Provided: {len(sn)}\nOutput Directory: {output_dir}/",
         "phase"
     )
-    return command
+    generate_conda_deps()
+    _module = subprocess.run(command)
+    sys.exit(_module.returncode)

src/harpy/popgroups.py

@@ -55,4 +55,3 @@ def popgroup(inputdir, output):
         for i in samplenames:
             _ = file.write(i + '\tpop1\n')
     print_notice(write_text + " Please review it, as all samples have been grouped into a single population")
-    return 0

src/harpy/preflight.py

@@ -2,7 +2,9 @@
 
 import os
 import sys
+import subprocess
 import rich_click as click
+from .conda_deps import generate_conda_deps
 from .helperfunctions import fetch_rule, fetch_report, fetch_script
 from .printfunctions import print_onstart
 from .fileparsers import parse_alignment_inputs, parse_fastq_inputs
@@ -86,7 +88,9 @@ def fastq(inputs, output_dir, threads, snakemake, quiet, conda, print_only):
         f"Files: {len(sn)}\nOutput Directory: {output_dir}/",
         "preflight fastq"
     )
-    return command
+    generate_conda_deps()
+    _module = subprocess.run(command)
+    sys.exit(_module.returncode)
 
 @click.command(no_args_is_help = True, epilog = "read the docs for more information: https://pdimens.github.io/harpy/modules/preflight/")
 @click.option('-t', '--threads', default = 4, show_default = True, type = click.IntRange(min = 1, max_open = True), help = 'Number of threads to use')
@@ -141,4 +145,6 @@ def bam(inputs, output_dir, threads, snakemake, quiet, conda, print_only):
         f"Samples: {len(sn)}\nOutput Directory: {output_dir}/",
         "preflight bam"
     )
-    return command
+    generate_conda_deps()
+    _module = subprocess.run(command)
+    sys.exit(_module.returncode)

src/harpy/qc.py

@@ -1,7 +1,9 @@
 """Harpy sequence adapter trimming and quality control"""
 
 import sys
+import subprocess
 import rich_click as click
+from .conda_deps import generate_conda_deps
 from .helperfunctions import fetch_report, fetch_rule, fetch_script
 from .fileparsers import get_samples_from_fastq, parse_fastq_inputs
 from .printfunctions import print_onstart
@@ -31,7 +33,7 @@
 @click.option('--conda',  is_flag = True, default = False, help = 'Use conda/mamba instead of container')
 @click.option('--skipreports',  is_flag = True, show_default = True, default = False, help = 'Don\'t generate HTML reports')
 @click.option('--print-only',  is_flag = True, hidden = True, show_default = True, default = False, help = 'Print the generated snakemake command and exit')
-@click.argument('input', required=True, type=click.Path(exists=True), nargs=-1)
+@click.argument('inputs', required=True, type=click.Path(exists=True), nargs=-1)
 def qc(inputs, output_dir, min_length, max_length, ignore_adapters, extra_params, threads, snakemake, skipreports, quiet, conda, print_only):
     """
     Remove adapters and quality trim sequences
@@ -69,7 +71,7 @@ def qc(inputs, output_dir, min_length, max_length, ignore_adapters, extra_params
     fetch_rule(workflowdir, "qc.smk")
     fetch_report(workflowdir, "BxCount.Rmd")
 
-    with open(f"{workflowdir}/config.yml", "w", encoding="uft-8") as config:
+    with open(f"{workflowdir}/config.yml", "w", encoding="utf-8") as config:
         config.write(f"seq_directory: {workflowdir}/input\n")
         config.write(f"output_directory: {output_dir}\n")
         config.write(f"adapters: {ignore_adapters}\n")
@@ -83,4 +85,6 @@ def qc(inputs, output_dir, min_length, max_length, ignore_adapters, extra_params
         f"Samples: {len(sn)}\nOutput Directory: {output_dir}/",
         "qc"
     )
-    return command
+    generate_conda_deps()
+    _module = subprocess.run(command)
+    sys.exit(_module.returncode)

src/harpy/simulate.py

@@ -2,7 +2,9 @@
 
 import os
 import sys
+import subprocess
 import rich_click as click
+from .conda_deps import generate_conda_deps
 from .helperfunctions import fetch_rule, fetch_script, symlink
 from .printfunctions import print_onstart, print_error
 from .validations import validate_input_by_ext
@@ -164,8 +166,9 @@ def linkedreads(genome_hap1, genome_hap2, output_dir, outer_distance, mutation_r
     onstart_text += f"Output Directory: {output_dir}/"
     print_onstart(onstart_text, "simulate reads")
 
-    return command
-
+    generate_conda_deps()
+    _module = subprocess.run(command)
+    sys.exit(_module.returncode)
 
 @click.command(no_args_is_help = True, epilog = "This workflow can be quite technical, please read the docs for more information: https://pdimens.github.io/harpy/modules/simulate/simulate-variants")
 @click.option('-s', '--snp-vcf', type=click.Path(exists=True, dir_okay=False), help = 'VCF file of known snps to simulate')
@@ -289,7 +292,9 @@ def snpindel(genome, snp_vcf, indel_vcf, output_dir, prefix, snp_count, indel_co
         config.write(f"workflow_call: {call_SM}\n")
 
     print_onstart(printmsg.rstrip("\n"), "simulate variants: snpindel")
-    return command
+    generate_conda_deps()
+    _module = subprocess.run(command)
+    sys.exit(_module.returncode)
 
 @click.command(no_args_is_help = True, epilog = "Please read the docs for more information: https://pdimens.github.io/harpy/modules/simulate/simulate-variants")
 @click.option('-v', '--vcf', type=click.Path(exists=True, dir_okay=False), help = 'VCF file of known inversions to simulate')
@@ -389,7 +394,9 @@ def inversion(genome, vcf, prefix, output_dir, count, min_size, max_size, centro
         config.write(f"workflow_call: {call_SM}\n")
 
     print_onstart(printmsg.rstrip("\n"), "simulate variants: inversion")
-    return command
+    generate_conda_deps()
+    _module = subprocess.run(command)
+    sys.exit(_module.returncode)
 
 @click.command(no_args_is_help = True, epilog = "Please read the docs for more information: https://pdimens.github.io/harpy/modules/simulate/simulate-variants")
 @click.option('-v', '--vcf', type=click.Path(exists=True, dir_okay=False), help = 'VCF file of known copy number variants to simulate')
@@ -501,7 +508,9 @@ def cnv(genome, output_dir, vcf, prefix, count, min_size, max_size, dup_ratio, m
         config.write(f"workflow_call: {call_SM}\n")
 
     print_onstart(printmsg.rstrip("\n"),"simulate cnv")
-    return command
+    generate_conda_deps()
+    _module = subprocess.run(command)
+    sys.exit(_module.returncode)
 
 @click.command(no_args_is_help = True, epilog = "Please read the docs for more information: https://pdimens.github.io/harpy/modules/simulate/simulate-variants")
 @click.option('-v', '--vcf', type=click.Path(exists=True, dir_okay=False), help = 'VCF file of known translocations to simulate')
@@ -597,4 +606,6 @@ def translocation(genome, output_dir, prefix, vcf, count, centromeres, genes, he
         config.write(f"workflow_call: {call_SM}\n")
 
     print_onstart(printmsg.rstrip("\n"), "simulate variants: translocation")
-    return command
+    generate_conda_deps()
+    _module = subprocess.run(command)
+    sys.exit(_module.returncode)

src/harpy/containerize.smk → src/harpy/snakefiles/containerize.smk

@@ -1,8 +1,5 @@
 import os
 import shutil
-from conda_deps import generate_conda_deps
-
-generate_conda_deps()
 
 onsuccess:
     shutil.rmtree(f'.harpy_envs', ignore_errors=True)