From 9409502620ea7f521a2035a241dfce24c9ab3900 Mon Sep 17 00:00:00 2001
From: pdimens <pdimens@live.com>
Date: Fri, 26 Apr 2024 09:59:47 -0400
Subject: [PATCH] update bcftools dep to 1.20 and update logic

---
 .github/filters.yml                          |  2 +-
 src/harpy/helperfunctions.py                 |  2 +-
 src/harpy/phase.py                           |  4 ++--
 src/harpy/rules/{phase-pop.smk => phase.smk} | 13 ++-----------
 4 files changed, 6 insertions(+), 15 deletions(-)
 rename src/harpy/rules/{phase-pop.smk => phase.smk} (97%)

diff --git a/.github/filters.yml b/.github/filters.yml
index 653ec4fa4..f44992fa7 100644
--- a/.github/filters.yml
+++ b/.github/filters.yml
@@ -106,7 +106,7 @@ naibr:
   - 'src/harpy/globalscripts/inferSV.py'
 phase:
   - *common
-  - 'src/harpy/rules/phase-pop.smk'
+  - 'src/harpy/rules/phase.smk'
   - '**phase.py'
   - 'test/bam/**'
   - 'test/vcf/test.bcf'
diff --git a/src/harpy/helperfunctions.py b/src/harpy/helperfunctions.py
index 2e85b1480..63a57453d 100644
--- a/src/harpy/helperfunctions.py
+++ b/src/harpy/helperfunctions.py
@@ -23,7 +23,7 @@ def generate_conda_deps():
     environ = {
         "qc" : ["falco", "fastp", "multiqc", "pysam=0.22"],
         "align": ["bwa", "ema","icu","libzlib", "minimap2", "samtools=1.19", "seqtk", "xz"],
-        "variants.snp": ["bcftools=1.19", "freebayes=1.3.6"],
+        "variants.snp": ["bcftools=1.20", "freebayes=1.3.6"],
         "variants.sv": ["leviathan", "naibr-plus"],
         "phase" : ["hapcut2", "whatshap"],
         "simulations" : ["perl", "perl-math-random", "perl-inline-c", "perl-parse-recdescent", "numpy", "dwgsim", "alienzj::msort"],
diff --git a/src/harpy/phase.py b/src/harpy/phase.py
index 85ca3bbae..abe71d996 100644
--- a/src/harpy/phase.py
+++ b/src/harpy/phase.py
@@ -38,7 +38,7 @@ def phase(input, output_dir, vcf, threads, molecule_distance, prune_threshold, v
     workflowdir = f"{output_dir}/workflow"
     command = f'snakemake --rerun-incomplete --rerun-triggers input mtime params --nolock --software-deployment-method conda --conda-prefix ./.snakemake/conda --cores {threads} --directory .'.split()
     command.append('--snakefile')
-    command.append(f"{workflowdir}/phase-pop.smk")
+    command.append(f"{workflowdir}/phase.smk")
     command.append("--configfile")
     command.append(f"{workflowdir}/config.yml")
     if quiet:
@@ -58,7 +58,7 @@ def phase(input, output_dir, vcf, threads, molecule_distance, prune_threshold, v
     validate_bamfiles(f"{workflowdir}/input", samplenames)
     if genome:
         validate_input_by_ext(genome, "--genome", [".fasta", ".fa", ".fasta.gz", ".fa.gz"])
-    fetch_rule(workflowdir, "phase-pop.smk")
+    fetch_rule(workflowdir, "phase.smk")
     fetch_report(workflowdir, "HapCut2.Rmd")
     prune_threshold /= 100
 
diff --git a/src/harpy/rules/phase-pop.smk b/src/harpy/rules/phase.smk
similarity index 97%
rename from src/harpy/rules/phase-pop.smk
rename to src/harpy/rules/phase.smk
index 75b6e80c3..c85f49fde 100644
--- a/src/harpy/rules/phase-pop.smk
+++ b/src/harpy/rules/phase.smk
@@ -219,21 +219,12 @@ rule merge_samples:
     output:
         bcf = outdir + "/variants.phased.bcf",
         idx = outdir + "/variants.phased.bcf.csi"
-    params:
-        "true" if len(samplenames) > 1 else "false"
     threads:
         30
     message:
         "Combining results into {output.bcf}" if len(samplenames) > 1 else "Copying results to {output.bcf}"
     shell:
-        """
-        if [ "{params}" = true ]; then
-            bcftools merge --threads {threads} -Ob -o {output.bcf} --write-index {input.bcf}
-        else
-           cp {input.bcf} {output.bcf}
-           cp {input.idx} {output.idx}
-        fi
-        """
+        "bcftools merge --threads {threads} -Ob -o {output.bcf} --force-single --write-index {input.bcf}"
 
 rule summarize_blocks:
     input:
@@ -292,7 +283,7 @@ rule log_workflow:
             _ = f.write("        awk '!/<ID=GX/' |\n")
             _ = f.write("        sed 's/:GX:/:GT:/' |\n")
             _ = f.write("        bcftools view -Ob -o sample.annot.bcf -\n")
-            _ = f.write("    bcftools merge --output-type b samples.annot.bcf\n\n")
+            _ = f.write("    bcftools merge --output-type b samples*.annot.bcf\n\n")
             _ = f.write("The header.file of extra vcf tags:\n")
             _ = f.write('    ##INFO=<ID=HAPCUT,Number=0,Type=Flag,Description="The haplotype was created with Hapcut2">\n')
             _ = f.write('    ##FORMAT=<ID=GX,Number=1,Type=String,Description="Haplotype">\n')