fix tmpdir being set from base command

aviary/aviary.py

@@ -35,7 +35,6 @@
 import logging
 import os
 from datetime import datetime
-import tempfile
 
 # Debug
 debug={1:logging.CRITICAL,
@@ -198,7 +197,7 @@ def main():
         help='Path to the location that will be treated used for temporary files. If none is specified, the TMPDIR \n'
              'environment variable will be used. Can be configured within the `configure` subcommand',
         dest='tmpdir',
-        default=tempfile.gettempdir(),
+        default=None,
     )
 
     base_group.add_argument(
@@ -1290,9 +1289,6 @@ def manage_env_vars(args):
     if args.conda_prefix is None:
         args.conda_prefix = Config.get_software_db_path('CONDA_ENV_PATH', '--conda-prefix')
 
-    if args.tmpdir is None:
-        args.tmpdir = Config.get_software_db_path('TMPDIR', '--tmpdir')
-
     try:
         if args.gtdb_path is None:
             args.gtdb_path = Config.get_software_db_path('GTDBTK_DATA_PATH', '--gtdb-path')

aviary/modules/annotation/annotation.smk

@@ -171,7 +171,7 @@ rule eggnog:
     params:
         mag_extension = config['mag_extension'],
         eggnog_db = config['eggnog_folder'],
-        tmpdir = config["tmpdir"]
+        tmpdir = config["tmpdir"] if config["tmpdir"] else "$TMPDIR",
     output:
         done = 'data/eggnog/done'
     threads:

aviary/modules/assembly/assembly.smk

@@ -362,7 +362,7 @@ rule spades_assembly:
         max_memory = config["max_memory"],
         long_read_type = config["long_read_type"],
         kmer_sizes = " ".join(config["kmer_sizes"]),
-        tmpdir = config["tmpdir"]
+        tmpdir = config["tmpdir"] if config["tmpdir"] else "$TMPDIR"
     conda:
         "envs/spades.yaml"
     benchmark:
@@ -458,14 +458,14 @@ rule spades_assembly_coverage:
     log:
         "logs/spades_assembly_coverage.log"
     params:
-         tmpdir = config["tmpdir"]
+         tmpdir = f"TMPDIR={config["tmpdir"]}" if config["tmpdir"] else ""
     conda:
          "../../envs/coverm.yaml"
     benchmark:
         "benchmarks/spades_assembly_coverage.benchmark.txt"
     shell:
         """
-        TMPDIR={params.tmpdir} coverm contig -m metabat -t {threads} -r {input.fasta} --interleaved {input.fastq} --bam-file-cache-directory data/cached_bams/ > {output.assembly_cov} 2> {log};
+        {params.tmpdir} coverm contig -m metabat -t {threads} -r {input.fasta} --interleaved {input.fastq} --bam-file-cache-directory data/cached_bams/ > {output.assembly_cov} 2> {log};
         mv data/cached_bams/*.bam {output.bam} && samtools index -@ {threads} {output.bam} 2>> {log}
         """
 

aviary/modules/assembly/scripts/assemble_short_reads.py

@@ -29,6 +29,15 @@ def assemble_short_reads(
     :return:
     '''
 
+    if not tmp_dir:
+        try:
+            tmp_dir = os.environ["TMPDIR"]
+        except KeyError:
+            tmp_dir_arg = ""
+
+    if tmp_dir:
+        tmp_dir_arg = f"--tmp-dir {tmp_dir}"
+
     # deal with read sets i.e. are we coassembling? Which assembler are we using?
     # Non co-assembled reads are handled the same for each assembler
     if read_set1 != 'none':
@@ -86,7 +95,7 @@ def assemble_short_reads(
     # Run chosen assembler
     if use_megahit:
         max_memory_in_bytes = max_memory * 1024*1024*1024
-        command = f"megahit {read_string} -t {threads} -m {max_memory_in_bytes} -o data/megahit_assembly --tmp-dir {tmp_dir}"
+        command = f"megahit {read_string} -t {threads} -m {max_memory_in_bytes} -o data/megahit_assembly {tmp_dir_arg}"
 
         with open(log, 'a') as logf:
             logf.write(f"Queueing command {command}\n")
@@ -97,7 +106,7 @@ def assemble_short_reads(
     else:
         kmers = " ".join(kmer_sizes)
         command = f"spades.py --memory {max_memory} --meta -t {threads} " \
-                f"-o data/short_read_assembly {read_string} -k {kmers} --tmp-dir {tmp_dir}"
+                f"-o data/short_read_assembly {read_string} -k {kmers} {tmp_dir_arg}"
         with open(log, 'a') as logf:
             logf.write(f"Queueing command {command}\n")
             subprocess.run(command.split(), stdout=logf, stderr=subprocess.STDOUT)

aviary/modules/binning/scripts/get_coverage.py

@@ -12,7 +12,8 @@ def get_coverage(
     threads: int,
     log: str,
 ):
-    os.environ["TMPDIR"] = tmpdir
+    if tmpdir: os.environ["TMPDIR"] = tmpdir
+
     if long_reads != "none" and not os.path.exists("data/long_cov.tsv"):
         if long_read_type in ["ont", "ont_hq"]:
             coverm_cmd = f"coverm contig -t {threads} -r {input_fasta} --single {' '.join(long_reads)} -p minimap2-ont -m length trimmed_mean variance --bam-file-cache-directory data/binning_bams/ --discard-unmapped --min-read-percent-identity 0.85 --output-file data/long_cov.tsv".split()

aviary/modules/processor.py

@@ -92,7 +92,7 @@ def __init__(self,
 
 
         self.conda_prefix = args.conda_prefix
-        self.tmpdir = os.path.abspath(args.tmpdir)
+        self.tmpdir = os.path.abspath(args.tmpdir) if args.tmpdir else None
         self.resources = args.resources
         self.output = os.path.abspath(args.output)
         self.threads = args.max_threads
@@ -444,7 +444,8 @@ def run_workflow(self, cores=16, profile=None, cluster_retries=None,
         self._validate_config()
 
         cores = max(int(self.threads), cores)
-        os.environ["TMPDIR"] = self.tmpdir
+        if self.tmpdir is not None:
+            os.environ["TMPDIR"] = self.tmpdir
         for workflow in self.workflows:
             cmd = (
                 "snakemake --snakefile {snakefile} --directory {working_dir} "

aviary/modules/strain_analysis/strain_analysis.smk

@@ -65,7 +65,6 @@ rule lorikeet:
     params:
         mag_extension = config['mag_extension'],
         parallel_genomes = 8,
-        tmpdir = config['tmpdir']
     resources:
         mem_mb=int(config["max_memory"])*512
     threads:

config.yaml

@@ -50,4 +50,4 @@ ani: none
 precluster_ani: none
 precluster_method: none
 pggb_params: none
-tmpdir: /tmp
+tmpdir: none