Update for compatibility with Snakemake 8+

workflow/Snakefile

@@ -136,14 +136,23 @@ if config["qc"]["rsq2"] and config["qc"]["rsq2"] != 'NA':
     qualfilt += " || (R2 >= {R2} && MAF < {MAF})".format(
         R2=config["qc"]["rsq2"], MAF=config["qc"]["maf"])
 
+
+if "fix_fam" in config and config["fix_fam"]:
+    path_plink_bycohort = "{impute_dir}/data/{cohort}_chrall_filtered_fixed.{ext}"
+    path_plink_merged = "{impute_dir}/data/all_chrall_filtered_fixed.{ext}"
+else:
+    path_plink_bycohort = "{impute_dir}/data/{cohort}_chrall_filtered.{ext}"
+    path_plink_merged = "{impute_dir}/data/all_chrall_filtered.{ext}"
+
+
 outs = dict(
     stat_report="{impute_dir}/stats/{cohort}_impStats.html",
     vcf_bycohort="{impute_dir}/data/{cohort}_chrall_filtered.vcf.gz",
     vcf_merged="{impute_dir}/data/all_chrall_filtered.vcf.gz",
     bgen_bycohort="{impute_dir}/data/{cohort}_chrall_filtered.bgen",
     bgen_merged="{impute_dir}/data/merged/merged_chrall_filtered.bgen",
-    plink_bycohort="{impute_dir}/data/{cohort}_chrall_filtered.{ext}",
-    plink_merged="{impute_dir}/data/all_chrall_filtered.{ext}")
+    plink_bycohort=path_plink_bycohort,
+    plink_merged=path_plink_merged)
 
 
 def expand_outs(out):
@@ -172,7 +181,7 @@ if zipped:
         conda: 'envs/p7z.yaml'
         threads: 4
         resources:
-            mem_mb = 4000,
+            mem_mb = 16000,
             time_min = 5
         shell:
             r'''
@@ -217,8 +226,8 @@ rule stats:
     conda: "envs/r_stats.yaml"
     threads: 22
     resources:
-        mem_mb = 8000,
-        walltime = '8:00'
+        mem = "256GB",
+        runtime = "8h"
     script: "scripts/Post_imputation.Rmd"
 
 # Sample filtering rules
@@ -244,7 +253,7 @@ rule fixheaders:
     threads: 1
     resources:
         mem_mb = 2048,
-        walltime = '24:00'
+        runtime = "24h"
     conda: "envs/bcftools.yaml"
     shell:
         r"""
@@ -270,7 +279,7 @@ if minimac_version == 'guess':
         threads: 1
         resources:
             mem_mb = 2048,
-            walltime = '24:00'
+            runtime = "24h"
         script: 'scripts/rule_detect_minimac.py'
 else:
     rule detect_minimac:
@@ -282,7 +291,7 @@ else:
         localrule: True
         resources:
             mem_mb = 2048,
-            walltime = '1:00'
+            runtime = "1h"
         shell: 'echo {params.ver} > {output}'
 
 
@@ -299,8 +308,8 @@ if sampfilt and minimac_version in ['guess', '3']:
             sf = sampfilt
         threads: 8
         resources:
-            mem_mb = 256,
-            walltime = '24:00'
+            mem_mb = 4096,
+            runtime = "24h"
         conda: "envs/bcftools.yaml"
         shell:
             r'''
@@ -332,8 +341,8 @@ elif minimac_version in ['guess', '3']:
             filt = qualfilt
         threads: 8
         resources:
-            mem_mb = 256,
-            walltime = '24:00'
+            mem_mb = 4096,
+            runtime = "24h"
         conda: "envs/bcftools.yaml"
         shell:
             r'''
@@ -364,8 +373,8 @@ elif sampfilt:
             sf = sampfilt
         threads: 8
         resources:
-            mem_mb = 256,
-            walltime = '24:00'
+            mem_mb = 4096,
+            runtime = "24h"
         conda: "envs/bcftools.yaml"
         shell:
             r'''
@@ -385,8 +394,8 @@ else:
             filt = qualfilt
         threads: 8
         resources:
-            mem_mb = 256,
-            walltime = '24:00'
+            mem_mb = 4096,
+            runtime = "24h"
         conda: "envs/bcftools.yaml"
         shell:
             r'''
@@ -429,7 +438,7 @@ rule rename:
     conda: "envs/bcftools.yaml"
     threads: 2
     resources:
-        mem_mb = 1024,
+        mem_mb = 2048,
         time_min = 60
     shell:
         '''
@@ -460,7 +469,7 @@ rule renameAuto:
     conda: "envs/bcftools.yaml"
     threads: 2
     resources:
-        mem_mb = 1024,
+        mem_mb = 2048,
         time_min = 60
     shell:
         '''
@@ -473,8 +482,8 @@ rule concat_chroms_samp:
     output: "{impute_dir}/data/{cohort}_chrall_filtered.vcf.gz"
     threads: 8
     resources:
-        mem_mb = 512,
-        walltime = '24:00'
+        mem_mb = 4096,
+        runtime = "24h"
     conda: "envs/bcftools.yaml"
     shell: "bcftools concat --threads 8 {input} | bcftools norm -d none -o {output} -Oz"
 
@@ -484,7 +493,7 @@ rule index_samples_chrom:
     conda: "envs/bcftools.yaml"
     threads: 1
     resources:
-        mem_mb = 256,
+        mem_mb = 1024,
         time_min = 120
     shell: "bcftools index -t {input}"
 
@@ -495,18 +504,31 @@ rule merge_samples_chrom:
     output: "{impute_dir}/data/by_chrom/all_chr{chrom}_filtered.vcf.gz"
     threads: 8
     resources:
-        mem_mb = 2000,
-        walltime = "36:00"
+        mem_mb = 16000,
+        runtime = "36h"
     conda: "envs/bcftools.yaml"
-    shell: "bcftools merge -m none --threads 8 {input.vcf} | bcftools norm -d none -o {output} -Oz"
+    shell:
+        """
+if [ $(echo {input.vcf} | wc -w) -eq 1 ]; then
+    cp {input.vcf} {output}
+else
+    bcftools merge -m none --threads 8 {input.vcf} | bcftools norm -d none -o {output} -Oz
+fi
+"""
+
+def tempif_cca(path):
+    if config.get("temp_ccs", False):
+        return temp(path)
+    else:
+        return path
 
 rule concat_chroms_all:
     input: expand("{{impute_dir}}/data/by_chrom/all_chr{chrom}_filtered.vcf.gz", chrom=CHROM)
-    output: "{impute_dir}/data/all_chrall_filtered.vcf.gz"
+    output: tempif_cca("{impute_dir}/data/all_chrall_filtered.vcf.gz")
     threads: 8
     resources:
-        mem_mb = 256,
-        walltime = "24:00"
+        mem_mb = 2048,
+        runtime =  "24h"
     conda: "envs/bcftools.yaml"
     shell: "bcftools concat -o {output} -Oz --threads 8 {input}"
 
@@ -518,8 +540,8 @@ rule make_plink_all:
         ID = "--id-delim" if automap_tf else "--double-id"
     threads: 10
     resources:
-        mem_mb = 3000,
-        walltime = "96:00"
+        mem_mb = 30000,
+        runtime = "96h"
     conda: "envs/plink.yaml"
     shell:
         "plink --keep-allele-order --vcf {input} {params.ID} --memory 20000 --threads 10 --make-bed "
@@ -533,12 +555,38 @@ rule make_plink_samp:
         ID = "--id-delim" if automap_tf else "--double-id "
     threads: 10
     resources:
-        mem_mb = 2000,
-        walltime = "2:00"
+        mem_mb = 20000,
+        runtime = "2h"
     conda: "envs/plink.yaml"
     shell:
         "plink --keep-allele-order --vcf {input} {params.ID} --memory 20000 --threads 10 --make-bed "
         "--out {params.out_plink}"
 
 # If bgen outputs are requested
 include: "rules/bgen.smk"
+
+if "fix_fam" in config and config["fix_fam"]:
+    rule fix_fam:
+        input:
+            oldfam = config["fix_fam"],
+            newfam = '{impute_dir}/data/{cohorts}_chrall_filtered.fam'
+        output: "{impute_dir}/data/{cohorts}_chrall_filtered_fixed.fam"
+        threads: 1
+        resources:
+            mem_mb = 1024,
+            time_min = 180
+        conda: "envs/r.yaml"
+        script: 'scripts/fix_fam.R'
+
+    rule link_fix_bedbim:
+        input:
+            bed = '{impute_dir}/data/{cohorts}_chrall_filtered.bed',
+            bim = '{impute_dir}/data/{cohorts}_chrall_filtered.bim'
+        output:
+            bed = '{impute_dir}/data/{cohorts}_chrall_filtered_fixed.bed',
+            bim = '{impute_dir}/data/{cohorts}_chrall_filtered_fixed.bim'
+        localrule: True
+        shell: '''
+    ln -rs {input.bed} {output.bed}
+    ln -rs {input.bim} {output.bim}
+    '''

workflow/envs/detect_minimac.yaml

@@ -1,5 +1,6 @@
 channels:
     - conda-forge
+    - bioconda
 dependencies:
     - python=3.11
-    - pysam=0.22.0
+    - pysam=0.23.1

workflow/rules/bgen.smk

@@ -8,7 +8,7 @@ rule make_bgen:
     threads: 1
     resources:
         mem_mb = 4000,
-        walltime = "24:00"
+        runtime = "24h"
     container: 'docker://befh/bgen:v1.1.7'
     shell:
         """
@@ -26,7 +26,7 @@ rule cat_bgen_samp:
     threads: 1
     resources:
         mem_mb = 5000,
-        walltime = "120:00"
+        runtime = "120h"
     container: 'docker://befh/bgen:v1.1.7'
     shell:
         """
@@ -47,8 +47,8 @@ rule make_bgen_allsamp:
         args = " ".join(["-g {} -s {}".format(gen, samp) for gen, samp in zip(bga_gen, bga_samp)])
     threads: 10
     resources:
-        mem_mb = 4000,
-        walltime = "24:00"
+        mem_mb = 40000,
+        runtime = "24h"
     container: 'docker://befh/bgen:v1.1.7'
     shell:
         """
@@ -65,6 +65,6 @@ rule cat_bgen_allsamp:
     threads: 1
     resources:
         mem_mb = 5000,
-        walltime = "120:00"
+        runtime = "120h"
     container: 'docker://befh/bgen:v1.1.7'
     shell: "cat-bgen -g {input.gen} -og {output.gen}; cp {input.samp} {output.samp}"