update config for snp

alsmith151 · Apr 28, 2024 · d448f8a · d448f8a
1 parent 282e155
commit d448f8a
Show file tree

Hide file tree

Showing 3 changed files with 104 additions and 34 deletions.
diff --git a/seqnado/config.py b/seqnado/config.py
@@ -7,7 +7,6 @@
 template_dir = os.path.join(package_dir, "workflow/config")
 
 
-# Helper Functions
 def get_user_input(prompt, default=None, is_boolean=False, choices=None):
     while True:
         user_input = (
@@ -143,19 +142,20 @@ def setup_configuration(assay, genome, template_data):
             )
 
     # Make bigwigs
-    template_data["make_bigwigs"] = get_user_input(
-        "Do you want to make bigwigs? (yes/no)", default="no", is_boolean=True
-    )
-    if template_data["make_bigwigs"]:
-        template_data["pileup_method"] = get_user_input(
-            "Pileup method:", default="deeptools", choices=["deeptools", "homer"]
-        )
-        template_data["scale"] = get_user_input(
-            "Scale bigwigs? (yes/no)", default="no", is_boolean=True
-        )
-        template_data["make_heatmaps"] = get_user_input(
-            "Do you want to make heatmaps? (yes/no)", default="no", is_boolean=True
+    if assay not in ["snp"]:
+        template_data["make_bigwigs"] = get_user_input(
+            "Do you want to make bigwigs? (yes/no)", default="no", is_boolean=True
         )
+        if template_data["make_bigwigs"]:
+            template_data["pileup_method"] = get_user_input(
+                "Pileup method:", default="deeptools", choices=["deeptools", "homer"]
+            )
+            template_data["scale"] = get_user_input(
+                "Scale bigwigs? (yes/no)", default="no", is_boolean=True
+            )
+            template_data["make_heatmaps"] = get_user_input(
+                "Do you want to make heatmaps? (yes/no)", default="no", is_boolean=True
+            )
 
     # Call peaks
     if assay in ["chip", "atac"]:
@@ -196,29 +196,75 @@ def setup_configuration(assay, genome, template_data):
         else "False"
     )
 
-    # Make UCSC hub
-    template_data["make_ucsc_hub"] = get_user_input(
-        "Do you want to make a UCSC hub? (yes/no)", default="no", is_boolean=True
+    # SNP options
+    template_data["call_snps"] = (
+        get_user_input("Call SNPs? (yes/no)", default="no", is_boolean=True)
+        if assay == "snp"
+        else "False"
+    )
+
+    template_data["snp_calling_method"] = (
+        get_user_input(
+            "SNP caller:",
+            default="bcftools",
+            choices=["bcftools", "deepvariant"],
+        )
+        if template_data["call_snps"]
+        else "False"
     )
 
-    template_data["UCSC_hub_directory"] = (
-        get_user_input("UCSC hub directory:", default="seqnado_output/hub/")
-        if template_data["make_ucsc_hub"]
-        else "seqnado_output/hub/"
+    template_data["fasta"] = (
+        get_user_input("Path to reference fasta:", default="path/to/reference.fasta")
+        if template_data["call_snps"]
+        else "False"
     )
-    template_data["email"] = (
-        get_user_input("What is your email address?", default=f"{username}@example.com")
-        if template_data["make_ucsc_hub"]
-        else f"{username}@example.com"
+    template_data["fasta_index"] = (
+        get_user_input(
+            "Path to reference fasta index:", default="path/to/reference.fasta.fai"
+        )
+        if template_data["call_snps"]
+        else "False"
     )
-    template_data["color_by"] = (
-        get_user_input("Color by (for UCSC hub):", default="samplename")
-        if template_data["make_ucsc_hub"]
-        else "samplename"
+
+    template_data["snp_database"] = (
+        get_user_input(
+            "Path to SNP database:",
+            default="path/to/snp_database",
+        )
+        if template_data["call_snps"]
+        else "False"
     )
 
+    # Make UCSC hub
+    if assay not in ["snp"]:
+        template_data["make_ucsc_hub"] = get_user_input(
+            "Do you want to make a UCSC hub? (yes/no)", default="no", is_boolean=True
+        )
+
+        template_data["UCSC_hub_directory"] = (
+            get_user_input("UCSC hub directory:", default="seqnado_output/hub/")
+            if template_data["make_ucsc_hub"]
+            else "seqnado_output/hub/"
+        )
+        template_data["email"] = (
+            get_user_input("What is your email address?", default=f"{username}@example.com")
+            if template_data["make_ucsc_hub"]
+            else f"{username}@example.com"
+        )
+        template_data["color_by"] = (
+            get_user_input("Color by (for UCSC hub):", default="samplename")
+            if template_data["make_ucsc_hub"]
+            else "samplename"
+        )
+
     template_data["options"] = (
-        TOOL_OPTIONS if assay in ["chip", "atac", "snp"] else TOOL_OPTIONS_RNA
+        TOOL_OPTIONS
+        if assay in ["chip", "atac"]
+        else (
+            TOOL_OPTIONS_RNA
+            if assay == "rna"
+            else TOOL_OPTIONS_SNP if assay == "snp" else ""
+        )
     )
 
 
@@ -300,6 +346,26 @@ def setup_configuration(assay, genome, template_data):
 """
 
 
+TOOL_OPTIONS_SNP = """
+trim_galore:
+    threads: 8
+    options: --2colour 20 
+
+bowtie2:
+    threads: 8
+    options:
+
+picard:
+    threads: 8
+    options:
+
+bcftools:
+    threads: 16
+    options:
+    
+"""
+
+
 def create_config(assay, genome, rerun, debug=False):
     env = Environment(loader=FileSystemLoader(template_dir), auto_reload=False)
 

diff --git a/seqnado/workflow/config/config.yaml.jinja b/seqnado/workflow/config/config.yaml.jinja
@@ -21,7 +21,6 @@ remove_blacklist: "{{remove_blacklist}}"
 blacklist: "{{blacklist}}"
 
 remove_pcr_duplicates_method: "{{remove_pcr_duplicates_method}}"
-library_complexity: "{{library_complexity}}"
 
 shift_atac_reads: "{{shift_atac_reads}}"
 
@@ -44,6 +43,11 @@ salmon_index: "{{salmon_index}}"
 
 run_deseq2: "{{run_deseq2}}"
 
+call_snps: "{{call_snps}}"
+snp_calling_method: "{{snp_calling_method}}"
+fasta: "{{fasta}}"
+fasta_index: "{{fasta_index}}"
+snp_database: "{{snp_database}}"
 
 make_ucsc_hub: "{{make_ucsc_hub}}"
 ucsc_hub_details:

diff --git a/seqnado/workflow/rules/variant.smk b/seqnado/workflow/rules/variant.smk
@@ -7,8 +7,8 @@ if config["call_snps"]:
         output:
             vcf="seqnado_output/variant/bcftools/{sample}.vcf.gz",
         params:
-            fasta=config["genome"]["fasta"],
-            faidx=config["genome"]["fasta_index"],
+            fasta=config["fasta"],
+            faidx=config["fasta_index"],
         resources:
             mem=lambda wildcards, attempt: f"{10 * 2 ** (attempt -1)}GB",
             runtime=lambda wildcards, attempt: f"{5 * 2 ** (attempt - 1)}h",
@@ -33,7 +33,7 @@ if config["call_snps"]:
         output:
             stats="seqnado_output/variant/bcftools/{sample}_filtered.stats.txt",
         params:
-            fasta=config["genome"]["fasta"],
+            fasta=config["fasta"],
         shell:
             "bcftools stats -F {params.fasta} -s - {input.vcf} > {output.stats}"
 
@@ -43,7 +43,7 @@ if config["call_snps"]:
         output:
             summary="seqnado_output/variant/bcftools/{sample}_summary.pdf",
         params:
-            fasta=config["genome"]["fasta"],
+            fasta=config["fasta"],
             out_dir="seqnado_output/variant/bcftools/",
         shell:
             "plot-vcfstats -p {params.out_dir} {input.stats}"