showing signs of life

Author: joshfactorial

neat/cli/commands/read_simulator.py

@@ -3,6 +3,7 @@
 """
 
 import argparse
+from pathlib import Path
 
 from ...read_simulator import read_simulator_runner
 from .base import BaseCommand

neat/read_simulator/parallel_runner.py

@@ -87,7 +87,7 @@ def main(options: Options) -> None:
         if options.produce_vcf:
             vcf = current_output_dir / options.vcf.name
 
-        current_options = Options(reference, current_output_dir, fq1, fq2, vcf, bam, options)
+        current_options = options.copy_with_changes(reference, current_output_dir, fq1, fq2, vcf, bam)
         output_opts.append(current_options)
 
     pool = Pool()

neat/read_simulator/runner.py

@@ -53,7 +53,7 @@ def read_simulator_runner(config: str, output_dir: str, file_prefix: str):
     # Read options file
     options = Options.from_cli(
         output_dir,
-        file_prefix,
+        str(file_prefix),
         config,
 
     )

neat/read_simulator/single_runner.py

@@ -128,7 +128,7 @@ def read_simulator_single(options: Options):
 
         local_bam_pickle_file = None
         if options.produce_bam:
-            local_bam_pickle_file = options.temp_dir_path / f'{options.output.stem}_tmp_{contig}_{threadidx}.p.gz'
+            local_bam_pickle_file = options.temp_dir_path / f'{options.output_dir}/{options.output_prefix}_tmp_{contig}_{threadidx}.p.gz'
 
         if threadidx == 1:
             # init_progress_info()

neat/read_simulator/utils/options.py

@@ -14,6 +14,7 @@
 trigger quick-run mode, setting most of the parameters to defaults.
 """
 import os
+from copy import deepcopy
 from typing import Any
 
 import numpy as np
@@ -50,7 +51,7 @@ class representing the options
 
     def __init__(self,
                  reference: Path | None = None,
-                 output_dir: Path | None = Path(os.getcwd()),
+                 output_dir: Path | None = Path.cwd(),
                  fq1: Path | None = None,
                  fq2: Path | None = None,
                  vcf: Path | None = None,
@@ -139,7 +140,7 @@ def __init__(self,
         super().__init__(**kwargs)
         self.reference: Path = reference
         self.output_dir: Path = output_dir
-        self.output = self.output_dir / output_prefix
+        self.output_prefix: str = output_prefix
         self.overwrite_output = overwrite_output
         self.rng_seed = rng_seed
         self.read_len: int = read_len
@@ -206,9 +207,6 @@ def from_cli(output_dir: Path,
         should be the highest acceptable value (inclusive).
         (type, default, criteria1 (low/'exists'), criteria2 (high/None)
         """
-        # TODO maybe redo as a dataclass?
-        output = output_dir / output_prefix
-
         # These are for checking
         defs = {
             'reference': (Path, None, 'exists', None),
@@ -331,6 +329,29 @@ def read_yaml(self, config_yaml: Path, args: dict):
                     value = Path(value)
                 args[key] = value
 
+    def copy_with_changes(self,
+                          reference: Path | None = None,
+                          current_output_dir: Path | None = None,
+                          fq1: Path | None = None,
+                          fq2: Path | None = None,
+                          vcf: Path | None = None,
+                          bam: Path | None = None,
+                          ):
+        return_options = deepcopy(self)
+        if reference is not None:
+            return_options.reference = reference
+        if current_output_dir is not None:
+            return_options.current_output_dir = current_output_dir
+        if fq1 is not None:
+            return_options.fq1 = fq1
+        if fq2 is not None:
+            return_options.fq2 = fq2
+        if vcf is not None:
+            return_options.vcf = vcf
+        if bam is not None:
+            return_options.bam = bam
+        return return_options
+
     def set_random_seed(self) -> Generator:
         """
         Sets up random number generator, which will be used for the run.
@@ -376,26 +397,26 @@ def log_configuration(self):
         files_to_produce = f'Producing the following files:\n'
         if self.produce_fastq:
             if self.paired_ended:
-                fq1 = f'{str(self.output)}_r1.fastq.gz'
-                fq2 = f'{str(self.output)}_r2.fastq.gz'
+                fq1 = f'{str(self.output_dir)}/{self.output_prefix}_r1.fastq.gz'
+                fq2 = f'{str(self.output_dir)}/{self.output_prefix}_r2.fastq.gz'
                 validate_output_path(fq1, True, self.overwrite_output)
                 validate_output_path(fq2, True, self.overwrite_output)
                 files_to_produce += f'\t- {fq1}\n'
                 files_to_produce += f'\t- {fq2}\n'
                 self.fq1 = Path(fq1)
                 self.fq2 = Path(fq2)
             else:
-                fq1 = f'{str(self.output)}.fastq.gz'
+                fq1 = f'{str(self.output_dir)}/{self.output_prefix}.fastq.gz'
                 validate_output_path(fq1, True, self.overwrite_output)
                 files_to_produce += f'\t- {fq1}\n'
                 self.fq1 = Path(fq1)
         if self.produce_bam:
-            bam = f'{self.output}_golden.bam'
+            bam = f'{str(self.output_dir)}/{self.output_prefix}_golden.bam'
             validate_output_path(bam, True, self.overwrite_output)
             files_to_produce += f'\t- {bam}\n'
             self.bam = Path(bam)
         if self.produce_vcf:
-            vcf = f'{self.output}_golden.vcf.gz'
+            vcf = f'{str(self.output_dir)}/{self.output_prefix}_golden.vcf.gz'
             validate_output_path(vcf, True, self.overwrite_output)
             files_to_produce += f'\t- {vcf}\n'
             self.vcf = Path(vcf)
@@ -420,8 +441,7 @@ def log_configuration(self):
                 if splits_dir.is_dir():
                     _LOG.info(f'Reusing existing splits {splits_dir}.')
                 else:
-                    _LOG.error(f'Reused splits set to True, but splits dir not found: {splits_dir}.')
-                    sys.exit(1)
+                    _LOG.warning(f'Reused splits set to True, but splits dir not found: {splits_dir}. Creating new splits')
 
             validate_output_path(splits_dir, False)
             self.splits_dir = splits_dir

neat/read_simulator/utils/output_file_writer.py

@@ -96,21 +96,21 @@ def __init__(self,
         # Set up filenames based on booleans
         files_to_write = []
         if self.paired and self.write_fastq:
-            self.fastq1_fn = options.output.parent / f'{options.output.stem}_r1.fastq.gz'
-            self.fastq2_fn = options.output.parent / f'{options.output.stem}_r2.fastq.gz'
+            self.fastq1_fn = options.fq1
+            self.fastq2_fn = options.fq2
             self.fastq_fns = [self.fastq1_fn, self.fastq2_fn]
             files_to_write.extend(self.fastq_fns)
         elif self.write_fastq:
-            self.fastq1_fn = options.output.parent / f'{options.output.stem}.fastq.gz'
+            self.fastq1_fn = options.fq1
             self.fastq2_fn = self.temporary_dir / "dummy.fastq.gz"
             self.fastq_fns = [self.fastq1_fn, self.fastq2_fn]
             files_to_write.extend(self.fastq_fns)
         if self.write_bam:
-            self.bam_fn = options.output.parent / f'{options.output.stem}_golden.bam'
+            self.bam_fn = options.bam
             self.bam_keys = list(bam_header)
             files_to_write.append(self.bam_fn)
         if self.write_vcf:
-            self.vcf_fn = options.output.parent / f'{options.output.stem}_golden.vcf.gz'
+            self.vcf_fn = options.vcf
             files_to_write.append(self.vcf_fn)
 
         self.files_to_write = files_to_write

neat/read_simulator/utils/stitch_outputs.py

@@ -20,7 +20,7 @@
 
 _LOG = logging.getLogger(__name__)
 
-def concat(files_to_join: List[(Path,)], dest: Path) -> None:
+def concat(files_to_join: List[Path], dest: Path) -> None:
     if not files_to_join:
         # Nothing to do, and no error to throw
         return