move to onstart to avoid double printing

src/harpy/snakefiles/align_bwa.smk

@@ -19,14 +19,16 @@ genome_zip  = True if bn.lower().endswith(".gz") else False
 bn_idx      = f"{bn}.gzi" if genome_zip else f"{bn}.fai"
 skipreports = config["skip_reports"]
 windowsize  = config["depth_windowsize"]
-os.makedirs(f"{outdir}/logs/snakemake", exist_ok = True)
-dt_string = datetime.now().strftime("%d_%m_%Y-%H_%M_%S")
-extra_logfile_handler = pylogging.FileHandler(f"{outdir}/logs/snakemake/{dt_string}.snakelog")
-logger.logger.addHandler(extra_logfile_handler)
 
 wildcard_constraints:
     sample = "[a-zA-Z0-9._-]+"
 
+onstart:
+    os.makedirs(f"{outdir}/logs/snakemake", exist_ok = True)
+    dt_string = datetime.now().strftime("%d_%m_%Y-%H_%M_%S")
+    extra_logfile_handler = pylogging.FileHandler(f"{outdir}/logs/snakemake/{dt_string}.snakelog")
+    logger.logger.addHandler(extra_logfile_handler)
+
 onerror:
     print("")
     rprint(

src/harpy/snakefiles/align_ema.smk

@@ -23,14 +23,16 @@ bn_idx      = f"{bn}.gzi" if genome_zip else f"{bn}.fai"
 envdir      = os.getcwd() + "/.harpy_envs"
 windowsize  = config["depth_windowsize"]
 skipreports = config["skip_reports"]
-os.makedirs(f"{outdir}/logs/snakemake", exist_ok = True)
-dt_string = datetime.now().strftime("%d_%m_%Y-%H_%M_%S")
-extra_logfile_handler = pylogging.FileHandler(f"{outdir}/logs/snakemake/{dt_string}.snakelog")
-logger.logger.addHandler(extra_logfile_handler)
 
 wildcard_constraints:
     sample = "[a-zA-Z0-9._-]+"
 
+onstart:
+    os.makedirs(f"{outdir}/logs/snakemake", exist_ok = True)
+    dt_string = datetime.now().strftime("%d_%m_%Y-%H_%M_%S")
+    extra_logfile_handler = pylogging.FileHandler(f"{outdir}/logs/snakemake/{dt_string}.snakelog")
+    logger.logger.addHandler(extra_logfile_handler)
+
 onerror:
     print("")
     rprint(

src/harpy/snakefiles/align_strobealign.smk

@@ -21,14 +21,16 @@ windowsize  = config["depth_windowsize"]
 molecule_distance = config["molecule_distance"]
 readlen = config["average_read_length"]
 autolen = isinstance(readlen, str)
-os.makedirs(f"{outdir}/logs/snakemake", exist_ok = True)
-dt_string = datetime.now().strftime("%d_%m_%Y-%H_%M_%S")
-extra_logfile_handler = pylogging.FileHandler(f"{outdir}/logs/snakemake/{dt_string}.snakelog")
-logger.logger.addHandler(extra_logfile_handler)
 
 wildcard_constraints:
     sample = "[a-zA-Z0-9._-]+"
 
+onstart:
+    os.makedirs(f"{outdir}/logs/snakemake", exist_ok = True)
+    dt_string = datetime.now().strftime("%d_%m_%Y-%H_%M_%S")
+    extra_logfile_handler = pylogging.FileHandler(f"{outdir}/logs/snakemake/{dt_string}.snakelog")
+    logger.logger.addHandler(extra_logfile_handler)
+
 onerror:
     print("")
     rprint(

src/harpy/snakefiles/deconvolve.smk

@@ -15,18 +15,19 @@ kmer_length = config["kmer_length"]
 window_size = config["window_size"]
 density 	= config["density"] 
 dropout     = config["dropout"]
-os.makedirs(f"{outdir}/logs/snakemake", exist_ok = True)
-dt_string = datetime.now().strftime("%d_%m_%Y-%H_%M_%S")
-extra_logfile_handler = pylogging.FileHandler(f"{outdir}/logs/snakemake/{dt_string}.snakelog")
-logger.logger.addHandler(extra_logfile_handler)
-
 
 bn_r = r"([_\.][12]|[_\.][FR]|[_\.]R[12](?:\_00[0-9])*)?\.((fastq|fq)(\.gz)?)$"
 samplenames = {re.sub(bn_r, "", os.path.basename(i), flags = re.IGNORECASE) for i in fqlist}
 
 wildcard_constraints:
     sample = "[a-zA-Z0-9._-]+"
 
+onstart:
+    os.makedirs(f"{outdir}/logs/snakemake", exist_ok = True)
+    dt_string = datetime.now().strftime("%d_%m_%Y-%H_%M_%S")
+    extra_logfile_handler = pylogging.FileHandler(f"{outdir}/logs/snakemake/{dt_string}.snakelog")
+    logger.logger.addHandler(extra_logfile_handler)
+
 onsuccess:
     print("")
     rprint(

src/harpy/snakefiles/demultiplex_gen1.smk

@@ -16,10 +16,6 @@ samplefile = config["inputs"]["demultiplex_schema"]
 skipreports = config["skip_reports"]
 outdir = config["output_directory"]
 envdir = os.getcwd() + "/.harpy_envs"
-os.makedirs(f"{outdir}/logs/snakemake", exist_ok = True)
-dt_string = datetime.now().strftime("%d_%m_%Y-%H_%M_%S")
-extra_logfile_handler = pylogging.FileHandler(f"{outdir}/logs/snakemake/{dt_string}.snakelog")
-logger.logger.addHandler(extra_logfile_handler)
 
 def barcodedict(smpl):
     d = {}
@@ -36,6 +32,12 @@ def barcodedict(smpl):
 samples = barcodedict(samplefile)
 samplenames = [i for i in samples.keys()]
 
+onstart:
+    os.makedirs(f"{outdir}/logs/snakemake", exist_ok = True)
+    dt_string = datetime.now().strftime("%d_%m_%Y-%H_%M_%S")
+    extra_logfile_handler = pylogging.FileHandler(f"{outdir}/logs/snakemake/{dt_string}.snakelog")
+    logger.logger.addHandler(extra_logfile_handler)
+
 onerror:
     print("")
     rprint(

src/harpy/snakefiles/impute.smk

@@ -20,11 +20,6 @@ skipreports = config["skip_reports"]
 outdir      = config["output_directory"]
 envdir      = os.getcwd() + "/.harpy_envs"
 paramspace  = Paramspace(pd.read_csv(paramfile, sep=r"\s+", skip_blank_lines=True).rename(columns=str.lower), param_sep = "", filename_params = ["k", "s", "ngen", "bxlimit"])
-os.makedirs(f"{outdir}/logs/snakemake", exist_ok = True)
-dt_string = datetime.now().strftime("%d_%m_%Y-%H_%M_%S")
-extra_logfile_handler = pylogging.FileHandler(f"{outdir}/logs/snakemake/{dt_string}.snakelog")
-logger.logger.addHandler(extra_logfile_handler)
-
 with open(biallelic, "r") as f_open:
     contigs = [i.rstrip() for i in f_open.readlines()]
 
@@ -36,6 +31,12 @@ def sam_index(infile):
     if not os.path.exists(f"{infile}.bai"):
         subprocess.run(f"samtools index {infile} {infile}.bai".split())
 
+onstart:
+    os.makedirs(f"{outdir}/logs/snakemake", exist_ok = True)
+    dt_string = datetime.now().strftime("%d_%m_%Y-%H_%M_%S")
+    extra_logfile_handler = pylogging.FileHandler(f"{outdir}/logs/snakemake/{dt_string}.snakelog")
+    logger.logger.addHandler(extra_logfile_handler)
+
 onerror:
     print("")
     rprint(

src/harpy/snakefiles/phase.smk

@@ -21,11 +21,6 @@ skipreports       = config["skip_reports"]
 samples_from_vcf  = config["samples_from_vcf"]
 variantfile       = config["inputs"]["variantfile"]
 bamlist     = config["inputs"]["alignments"]
-os.makedirs(f"{outdir}/logs/snakemake", exist_ok = True)
-dt_string = datetime.now().strftime("%d_%m_%Y-%H_%M_%S")
-extra_logfile_handler = pylogging.FileHandler(f"{outdir}/logs/snakemake/{dt_string}.snakelog")
-logger.logger.addHandler(extra_logfile_handler)
-
 # toggle linked-read aware mode
 if config["ignore_bx"]:
     fragfile = outdir + "/extract_hairs/{sample}.unlinked.frags"
@@ -40,7 +35,6 @@ if samples_from_vcf:
 else:
     samplenames = [Path(i).stem for i in bamlist]
 
-
 # toggle indel mode
 if config["inputs"].get("genome", None):
     genomefile = config["inputs"]["genome"]
@@ -63,6 +57,12 @@ else:
 wildcard_constraints:
     sample = "[a-zA-Z0-9._-]+"
 
+onstart:
+    os.makedirs(f"{outdir}/logs/snakemake", exist_ok = True)
+    dt_string = datetime.now().strftime("%d_%m_%Y-%H_%M_%S")
+    extra_logfile_handler = pylogging.FileHandler(f"{outdir}/logs/snakemake/{dt_string}.snakelog")
+    logger.logger.addHandler(extra_logfile_handler)
+
 onerror:
     print("")
     rprint(

src/harpy/snakefiles/preflight_bam.smk

@@ -14,15 +14,17 @@ from rich.panel import Panel
 outdir = config["output_directory"]
 envdir  = os.getcwd() + "/.harpy_envs"
 bamlist = config["inputs"]
-samplenames = {Path(i).stem for i in bamlist}  
-os.makedirs(f"{outdir}/logs/snakemake", exist_ok = True)
-dt_string = datetime.now().strftime("%d_%m_%Y-%H_%M_%S")
-extra_logfile_handler = pylogging.FileHandler(f"{outdir}/logs/snakemake/{dt_string}.snakelog")
-logger.logger.addHandler(extra_logfile_handler)
+samplenames = {Path(i).stem for i in bamlist}
 
 wildcard_constraints:
     sample = "[a-zA-Z0-9._-]+"
 
+onstart:
+    os.makedirs(f"{outdir}/logs/snakemake", exist_ok = True)
+    dt_string = datetime.now().strftime("%d_%m_%Y-%H_%M_%S")
+    extra_logfile_handler = pylogging.FileHandler(f"{outdir}/logs/snakemake/{dt_string}.snakelog")
+    logger.logger.addHandler(extra_logfile_handler)
+
 onerror:
     print("")
     rprint(

src/harpy/snakefiles/preflight_fastq.smk

@@ -15,14 +15,16 @@ outdir = config["output_directory"]
 envdir      = os.getcwd() + "/.harpy_envs"
 bn_r = r"([_\.][12]|[_\.][FR]|[_\.]R[12](?:\_00[0-9])*)?\.((fastq|fq)(\.gz)?)$"
 samplenames = {re.sub(bn_r, "", os.path.basename(i), flags = re.IGNORECASE) for i in fqlist}
-os.makedirs(f"{outdir}/logs/snakemake", exist_ok = True)
-dt_string = datetime.now().strftime("%d_%m_%Y-%H_%M_%S")
-extra_logfile_handler = pylogging.FileHandler(f"{outdir}/logs/snakemake/{dt_string}.snakelog")
-logger.logger.addHandler(extra_logfile_handler)
 
 wildcard_constraints:
     sample = "[a-zA-Z0-9._-]+"
 
+onstart:
+    os.makedirs(f"{outdir}/logs/snakemake", exist_ok = True)
+    dt_string = datetime.now().strftime("%d_%m_%Y-%H_%M_%S")
+    extra_logfile_handler = pylogging.FileHandler(f"{outdir}/logs/snakemake/{dt_string}.snakelog")
+    logger.logger.addHandler(extra_logfile_handler)
+
 onerror:
     print("")
     rprint(

src/harpy/snakefiles/qc.smk

@@ -25,17 +25,19 @@ if deconvolve:
     decon_d = deconvolve["density"]
     decon_a = deconvolve["dropout"]
 skipreports  = config["skip_reports"]
-os.makedirs(f"{outdir}/logs/snakemake", exist_ok = True)
-dt_string = datetime.now().strftime("%d_%m_%Y-%H_%M_%S")
-extra_logfile_handler = pylogging.FileHandler(f"{outdir}/logs/snakemake/{dt_string}.snakelog")
-logger.logger.addHandler(extra_logfile_handler)
 
 bn_r = r"([_\.][12]|[_\.][FR]|[_\.]R[12](?:\_00[0-9])*)?\.((fastq|fq)(\.gz)?)$"
 samplenames = {re.sub(bn_r, "", os.path.basename(i), flags = re.IGNORECASE) for i in fqlist}
 
 wildcard_constraints:
     sample = "[a-zA-Z0-9._-]+"
 
+onstart:
+    os.makedirs(f"{outdir}/logs/snakemake", exist_ok = True)
+    dt_string = datetime.now().strftime("%d_%m_%Y-%H_%M_%S")
+    extra_logfile_handler = pylogging.FileHandler(f"{outdir}/logs/snakemake/{dt_string}.snakelog")
+    logger.logger.addHandler(extra_logfile_handler)
+
 onsuccess:
     print("")
     rprint(

src/harpy/snakefiles/simulate_linkedreads.smk

@@ -19,11 +19,12 @@ if barcodes:
     barcodefile = barcodes
 else:
     barcodefile = f"{outdir}/workflow/input/4M-with-alts-february-2016.txt"
-os.makedirs(f"{outdir}/logs/snakemake", exist_ok = True)
-dt_string = datetime.now().strftime("%d_%m_%Y-%H_%M_%S")
-extra_logfile_handler = pylogging.FileHandler(f"{outdir}/logs/snakemake/{dt_string}.snakelog")
-logger.logger.addHandler(extra_logfile_handler)
 
+onstart:
+    os.makedirs(f"{outdir}/logs/snakemake", exist_ok = True)
+    dt_string = datetime.now().strftime("%d_%m_%Y-%H_%M_%S")
+    extra_logfile_handler = pylogging.FileHandler(f"{outdir}/logs/snakemake/{dt_string}.snakelog")
+    logger.logger.addHandler(extra_logfile_handler)
 
 onsuccess:
     shutil.rmtree('./_Inline', ignore_errors=True)

src/harpy/snakefiles/simulate_snpindel.smk

@@ -15,11 +15,6 @@ snp_vcf = config["inputs"].get("snp_vcf", None)
 indel_vcf = config["inputs"].get("indel_vcf", None)
 heterozygosity = config["heterozygosity"]
 outprefix = config["prefix"]
-os.makedirs(f"{outdir}/logs/snakemake", exist_ok = True)
-dt_string = datetime.now().strftime("%d_%m_%Y-%H_%M_%S")
-extra_logfile_handler = pylogging.FileHandler(f"{outdir}/logs/snakemake/{dt_string}.snakelog")
-logger.logger.addHandler(extra_logfile_handler)
-
 in_vcfs = []
 snp = False 
 indel = False
@@ -65,6 +60,12 @@ else:
 
 variants = [i for i,j in zip(["snp", "indel"], [snp, indel]) if j]
 
+onstart:
+    os.makedirs(f"{outdir}/logs/snakemake", exist_ok = True)
+    dt_string = datetime.now().strftime("%d_%m_%Y-%H_%M_%S")
+    extra_logfile_handler = pylogging.FileHandler(f"{outdir}/logs/snakemake/{dt_string}.snakelog")
+    logger.logger.addHandler(extra_logfile_handler)
+
 onsuccess:
     print("")
     rprint(

src/harpy/snakefiles/simulate_variants.smk

@@ -16,11 +16,6 @@ genome = config["inputs"]["genome"]
 vcf = config["inputs"].get("vcf", None)
 heterozygosity = float(config["heterozygosity"])
 vcf_correct = "None"
-os.makedirs(f"{outdir}/logs/snakemake", exist_ok = True)
-dt_string = datetime.now().strftime("%d_%m_%Y-%H_%M_%S")
-extra_logfile_handler = pylogging.FileHandler(f"{outdir}/logs/snakemake/{dt_string}.snakelog")
-logger.logger.addHandler(extra_logfile_handler)
-
 if vcf:
     vcf_correct = vcf[:-4] + ".vcf.gz" if vcf.lower().endswith("bcf") else vcf
     variant_params = f"-{variant}_vcf {vcf_correct}"
@@ -52,6 +47,12 @@ else:
         cnv_ratio =config.get("cnv_ratio", None)
         variant_params += f" --cnv_gain_loss_ratio {cnv_ratio}" if cnv_ratio else ""
 
+onstart:
+    os.makedirs(f"{outdir}/logs/snakemake", exist_ok = True)
+    dt_string = datetime.now().strftime("%d_%m_%Y-%H_%M_%S")
+    extra_logfile_handler = pylogging.FileHandler(f"{outdir}/logs/snakemake/{dt_string}.snakelog")
+    logger.logger.addHandler(extra_logfile_handler)
+
 onsuccess:
     print("")
     rprint(

src/harpy/snakefiles/snp_freebayes.smk

@@ -42,14 +42,16 @@ else:
             cont,startpos,endpos = line.split()
             intervals.add(f"{cont}:{startpos}-{endpos}")
     regions = dict(zip(intervals, intervals))
-os.makedirs(f"{outdir}/logs/snakemake", exist_ok = True)
-dt_string = datetime.now().strftime("%d_%m_%Y-%H_%M_%S")
-extra_logfile_handler = pylogging.FileHandler(f"{outdir}/logs/snakemake/{dt_string}.snakelog")
-logger.logger.addHandler(extra_logfile_handler)
 
 wildcard_constraints:
     sample = "[a-zA-Z0-9._-]+"
 
+onstart:
+    os.makedirs(f"{outdir}/logs/snakemake", exist_ok = True)
+    dt_string = datetime.now().strftime("%d_%m_%Y-%H_%M_%S")
+    extra_logfile_handler = pylogging.FileHandler(f"{outdir}/logs/snakemake/{dt_string}.snakelog")
+    logger.logger.addHandler(extra_logfile_handler)
+
 onerror:
     print("")
     rprint(

src/harpy/snakefiles/snp_mpileup.smk

@@ -37,14 +37,16 @@ else:
             cont,startpos,endpos = line.split()
             intervals.add(f"{cont}:{startpos}-{endpos}")
     regions = dict(zip(intervals, intervals))
-os.makedirs(f"{outdir}/logs/snakemake", exist_ok = True)
-dt_string = datetime.now().strftime("%d_%m_%Y-%H_%M_%S")
-extra_logfile_handler = pylogging.FileHandler(f"{outdir}/logs/snakemake/{dt_string}.snakelog")
-logger.logger.addHandler(extra_logfile_handler)
 
 wildcard_constraints:
     sample = "[a-zA-Z0-9._-]+"
 
+onstart:
+    os.makedirs(f"{outdir}/logs/snakemake", exist_ok = True)
+    dt_string = datetime.now().strftime("%d_%m_%Y-%H_%M_%S")
+    extra_logfile_handler = pylogging.FileHandler(f"{outdir}/logs/snakemake/{dt_string}.snakelog")
+    logger.logger.addHandler(extra_logfile_handler)
+
 onerror:
     print("")
     rprint(

src/harpy/snakefiles/sv_leviathan.smk

@@ -24,14 +24,16 @@ bn          = os.path.basename(genomefile)
 genome_zip  = True if bn.lower().endswith(".gz") else False
 if genome_zip:
     bn = bn[:-3]
-os.makedirs(f"{outdir}/logs/snakemake", exist_ok = True)
-dt_string = datetime.now().strftime("%d_%m_%Y-%H_%M_%S")
-extra_logfile_handler = pylogging.FileHandler(f"{outdir}/logs/snakemake/{dt_string}.snakelog")
-logger.logger.addHandler(extra_logfile_handler)
 
 wildcard_constraints:
     sample = "[a-zA-Z0-9._-]+"
 
+onstart:
+    os.makedirs(f"{outdir}/logs/snakemake", exist_ok = True)
+    dt_string = datetime.now().strftime("%d_%m_%Y-%H_%M_%S")
+    extra_logfile_handler = pylogging.FileHandler(f"{outdir}/logs/snakemake/{dt_string}.snakelog")
+    logger.logger.addHandler(extra_logfile_handler)
+
 onerror:
     print("")
     rprint(

src/harpy/snakefiles/sv_leviathan_pop.smk

@@ -21,15 +21,17 @@ skipreports = config["skip_reports"]
 bn 			= os.path.basename(genomefile)
 if bn.lower().endswith(".gz"):
     bn = bn[:-3]
-os.makedirs(f"{outdir}/logs/snakemake", exist_ok = True)
-dt_string = datetime.now().strftime("%d_%m_%Y-%H_%M_%S")
-extra_logfile_handler = pylogging.FileHandler(f"{outdir}/logs/snakemake/{dt_string}.snakelog")
-logger.logger.addHandler(extra_logfile_handler)
 
 wildcard_constraints:
     sample = "[a-zA-Z0-9._-]+",
     population = "[a-zA-Z0-9._-]+"
 
+onstart:
+    os.makedirs(f"{outdir}/logs/snakemake", exist_ok = True)
+    dt_string = datetime.now().strftime("%d_%m_%Y-%H_%M_%S")
+    extra_logfile_handler = pylogging.FileHandler(f"{outdir}/logs/snakemake/{dt_string}.snakelog")
+    logger.logger.addHandler(extra_logfile_handler)
+
 onerror:
     print("")
     rprint(