das_tool: Use script not shell.

aviary/modules/binning/binning.smk

@@ -629,32 +629,15 @@ rule das_tool:
         mem_mb = lambda wildcards, attempt: min(int(config["max_memory"])*1024, 512*1024*attempt),
         runtime = lambda wildcards, attempt: 12*60*attempt,
     output:
-        das_tool_done = "data/das_tool_bins_pre_refine/done"
+        touch("data/das_tool_bins_pre_refine/done")
     conda:
         "envs/das_tool.yaml"
     log:
         "logs/das_tool.log"
     benchmark:
         "benchmarks/das_tool.benchmark.txt"
-    shell:
-        """
-        Fasta_to_Scaffolds2Bin.sh -i data/metabat_bins_sspec -e fa > data/metabat_bins_sspec.tsv 2> {log}; 
-        Fasta_to_Scaffolds2Bin.sh -i data/metabat_bins_ssens -e fa > data/metabat_bins_ssens.tsv 2>> {log}; 
-        Fasta_to_Scaffolds2Bin.sh -i data/metabat_bins_sens -e fa > data/metabat_bins_sens.tsv 2>> {log}; 
-        Fasta_to_Scaffolds2Bin.sh -i data/metabat_bins_spec -e fa > data/metabat_bins_spec.tsv 2>> {log}; 
-        Fasta_to_Scaffolds2Bin.sh -i data/concoct_bins -e fa > data/concoct_bins.tsv 2>> {log}; 
-        Fasta_to_Scaffolds2Bin.sh -i data/maxbin2_bins -e fasta > data/maxbin_bins.tsv 2>> {log}; 
-        Fasta_to_Scaffolds2Bin.sh -i data/vamb_bins/bins -e fna > data/vamb_bins.tsv 2>> {log}; 
-        Fasta_to_Scaffolds2Bin.sh -i data/rosella_refined/final_bins/ -e fna > data/rosella_refined_bins.tsv 2>> {log}; 
-        Fasta_to_Scaffolds2Bin.sh -i data/metabat2_refined/final_bins/ -e fna > data/metabat2_refined_bins.tsv 2>> {log}; 
-        Fasta_to_Scaffolds2Bin.sh -i data/semibin_refined/final_bins/ -e fna > data/semibin_refined_bins.tsv 2>> {log}; 
-        scaffold2bin_files=$(find data/*bins*.tsv -not -empty -exec ls {{}} \; | tr "\n" ',' | sed "s/,$//g"); 
-        DAS_Tool --search_engine diamond --write_bin_evals 1 --write_bins 1 -t {threads} --score_threshold -42 \
-         -i $scaffold2bin_files \
-         -c {input.fasta} \
-         -o data/das_tool_bins_pre_refine/das_tool >> {log} 2>&1 && \
-        touch data/das_tool_bins_pre_refine/done
-        """
+    script:
+        "scripts/das_tool.py"
 
 rule refine_dastool:
     input:

aviary/modules/binning/envs/das_tool.yaml

@@ -3,3 +3,4 @@ channels:
   - bioconda
 dependencies:
   - das_tool = 1.1.2
+  - extern = 0.4.1

aviary/modules/binning/scripts/das_tool.py

@@ -0,0 +1,61 @@
+import logging
+import os
+import sys
+
+import extern
+
+if __name__ == '__main__':
+    unrefined_binners_to_use = [
+        ('concoct', 'fa'),
+        ('maxbin2', 'fasta'),
+        ('vamb', 'fna')]
+    refined_binners_to_use = [
+        ('rosella', 'fna'),
+        ('semibin', 'fna')]
+
+    # N.B. specifying "metabat" will skip both MetaBAT1 and MetaBAT2.
+    metabats = ['metabat_sspec', 'metabat_ssens', 'metabat_sens', 'metabat_spec']
+
+    binners = []
+    for (binner, extension) in unrefined_binners_to_use:
+        if binner not in snakemake.config['skip_binners']:
+            binners.append((f'{binner}_bins/', extension, f'data/{binner}_bins.tsv'))
+
+    for (binner, extension) in refined_binners_to_use:
+        if binner not in snakemake.config['skip_binners']:
+            binners.append((binner+'_refined/final_bins/', extension, f'data/{binner}_refined_bins.tsv'))
+
+    for metabat in metabats:
+        if metabat not in snakemake.config['skip_binners']:
+            binners.append((metabat.replace('metabat','metabat_bins'), 'fa', f'data/{metabat}_bins.tsv'))
+    if 'metabat2' not in snakemake.config['skip_binners']:
+        binners.append(('metabat2_refined/final_bins/', 'fna', f'data/metabat2_refined_bins.tsv'))
+
+    logfile = snakemake.log[0]
+    logging.basicConfig(filename=logfile, level=logging.INFO)
+    logging.info("Using the following binners: " + str(binners))
+
+    if len(binners) == 0:
+        logging.error("All binners have been skipped, so DAS_tool cannot be run.")
+        sys.exit(1)
+    if len(binners) == 1:
+        logging.error("Only one binner has been specified, so DAS_tool cannot be run. Please specify at least two binners.")
+        sys.exit(1)
+
+    bin_definition_files = []
+    for binner, extension, bin_definition_file in binners:
+        extern.run(f'Fasta_to_Scaffolds2Bin.sh -i data/{binner} -e {extension} >{bin_definition_file}  2>> {logfile}')
+        if os.path.getsize(bin_definition_file) == 0:
+            logging.warning(f'Bin definition file {bin_definition_file} is empty, suggesting that {binner} failed or did not not create any output bins.')
+        else:
+            bin_definition_files.append(bin_definition_file)
+    logging.info("Bin definition files created: " + str(bin_definition_files))
+
+    scaffold2bin_files = ','.join(bin_definition_files)
+
+    das_tool_command = f'DAS_Tool --search_engine diamond --write_bin_evals 1 --write_bins 1 -t {snakemake.threads} --score_threshold -42 \
+        -i {scaffold2bin_files} \
+        -c {snakemake.input.fasta} \
+        -o data/das_tool_bins_pre_refine/das_tool >> {logfile} 2>&1'
+    logging.info("Running DAS_Tool with command: " + das_tool_command)
+    extern.run(das_tool_command)

test/test_integration.py

@@ -140,7 +140,7 @@ def test_short_read_recovery_fast(self):
             f"-1 {data}/wgsim.1.fq.gz "
             f"-2 {data}/wgsim.2.fq.gz "
             f"--skip-abundances "
-            f"--skip-binners concoct rosella vamb metabat maxbin "
+            f"--skip-binners concoct rosella vamb maxbin2 semibin "
             f"--skip-qc "
             f"--refinery-max-iterations 0 "
             f"--conda-prefix {path_to_conda} "