refactor: require FASTA index as input for all-sites VCF

build: update bcftools version for all-sites VCF

Author: ahmig

workflow/envs/var_calling.yaml

@@ -4,4 +4,3 @@ channels:
 dependencies:
   - ivar==1.4.2
   - samtools==1.17
-  - bcftools==1.17

workflow/rules/fasta.smk

@@ -17,6 +17,18 @@ rule read_bam_refs:
         """
 
 
+rule index_fasta:
+    conda:
+        "../envs/tools.yaml"
+    input:
+        fasta="{name}.fasta",
+    output:
+        fai=temp("{name}.fasta.fai"),
+        sizes=temp("{name}.sizes.txt"),
+    shell:
+        "samtools faidx {input.fasta} -o {output.fai} && cut -f1,2 {output.fai} >{output.sizes}"
+
+
 rule rename_fastas:
     input:
         fasta = get_input_fasta

workflow/rules/sites.smk

@@ -18,14 +18,15 @@ rule problematic_vcf_to_bed:
 
 rule bcftools_mpileup_all_sites:
     threads: 1
-    conda: "../envs/var_calling.yaml"
+    conda: "../envs/tools.yaml"
     params:
         min_mq = 0,
         min_bq = config["VC"]["MIN_QUALITY"],
         mpileup_extra = "--no-BAQ"
     input:
         bam = get_input_bam,
         reference = OUTDIR/"vaf"/"{sample}.reference.fasta",
+        reference_index = OUTDIR/"vaf"/"{sample}.reference.fasta.fai",
     output:
         mpileup = temp(OUTDIR / "all_sites" / "{sample}.mpileup.vcf"),
         query = temp(OUTDIR / "all_sites" / "{sample}.query.tsv"),