Merge pull request #6 from cokelaer/master

Fixing problem when there is no annotation

.github/workflows/main.yml

@@ -13,7 +13,7 @@ jobs:
     strategy:
       max-parallel: 5
       matrix:
-        python: [3.7,3.8]
+        python: [3.7,3.8,3.9]
       fail-fast: false
 
 
@@ -40,7 +40,7 @@ jobs:
     - name: conda
       run: |
         conda install -c conda-forge mamba --quiet
-        mamba install -c bioconda -c conda-forge --quiet -y freebayes bwa snpeff sambamba picard samtools
+        mamba install -c bioconda -c conda-forge --quiet -y freebayes bwa snpeff==5.0 sambamba picard samtools bamtools
 
     - name: Install dependencies
       run: |

sequana_pipelines/variant_calling/main.py

@@ -124,14 +124,13 @@ def main(args=None):
                 logger.error("The annotation file must end with .gbk or .gff or .gff3. You provided {}".format(cfg.annotation_file))
                 sys.exit(1)
             cfg['sequana_coverage']['genbank_file'] = cfg.annotation_file
+        else:
+            cfg.snpeff.do = False
+            cfg['sequana_coverage']['genbank_file'] = ""
 
         cfg['sequana_coverage']['do'] = options.do_coverage
         cfg['sequana_coverage']["circular"] = options.circular
-
-
         cfg['joint_freebayes']['do'] = options.do_joint_calling
-
-
         cfg['bwa_mem']['threads'] = options.threads
         cfg['freebayes']['ploidy'] = options.freebayes_ploidy
 

sequana_pipelines/variant_calling/requirements.txt

@@ -4,3 +4,4 @@
 - sambamba
 - samtools
 - snpEff
+- bamtools

sequana_pipelines/variant_calling/variant_calling.rules

@@ -33,7 +33,6 @@ configfile: "config.yaml"
 
 # A convenient manager
 manager = PipelineManager("variant_calling", config)
-manager.setup(globals(), mode="warning")
 
 config = manager.config
 
@@ -48,9 +47,8 @@ if len(manager.samples) == 1:
 
 
 # ================================================== Define outputs
+expected_output = []
 
-__freebayes__output = "{sample}/freebayes/{sample}.raw.vcf"
-expected_output += expand(__freebayes__output, sample=manager.samples)
 
 others = []
 if config['joint_freebayes']['do']:
@@ -227,8 +225,7 @@ if config["sequana_coverage"]["do"]:
     rule sequana_coverage:
         input: 
             bed="{sample}/samtools_depth/{sample}.bed",
-            fasta=reference_file,
-            gbk=annotation_file if config["snpeff"]["do"] else ""
+            fasta=reference_file
         output:
             "{sample}/sequana_coverage/sequana_coverage.html"
         params:
@@ -239,11 +236,11 @@ if config["sequana_coverage"]["do"]:
             gc_window_size=config["sequana_coverage"]["gc_window_size"],
             high_threshold=config["sequana_coverage"]["high_threshold"],
             low_threshold=config["sequana_coverage"]["low_threshold"],
-            mixture_models=config["sequana_coverage"]["mixture_models"]
+            mixture_models=config["sequana_coverage"]["mixture_models"],
+            gbk=annotation_file if config["snpeff"]["do"] else None
         wrapper:
             f"{sequana_wrapper_branch}/wrappers/sequana_coverage"
 
-    #__sequana_coverage__report_dir = "{sample}/sequana_coverage"
     expected_output += expand("{sample}/sequana_coverage/sequana_coverage.html",
              sample=manager.samples)
 
@@ -274,7 +271,7 @@ if config["snpeff"]["do"]:
 
     rule snpeff:
         input:
-            vcf = __freebayes__output,
+            vcf = "{sample}/freebayes/{sample}.raw.vcf",
             ann = annotation_file
         output:
             html="{sample}/snpeff/{sample}.snpeff.html",
@@ -291,9 +288,13 @@ if config["snpeff"]["do"]:
     __freebayes_vcf_filter__input = "{sample}/snpeff/{sample}.ann.vcf"
     expected_output += expand("{sample}/snpeff/{sample}.ann.vcf", sample=manager.samples)
 else:
-    __freebayes_vcf_filter__input = __freebayes__output
+    __freebayes_vcf_filter__input = "{sample}/freebayes/{sample}.raw.vcf"
 
-# Freebayes filter
+expected_output += expand("{sample}/freebayes/{sample}.raw.vcf", sample=manager.samples)
+
+
+# ================================================================== Freebayes filter
+#
 #
 rule freebayes_vcf_filter:
     input:

test/test_main.py

@@ -20,7 +20,7 @@ def test_standalone_subprocess():
     subprocess.call(cmd.split())
 
 
-def test_check_output():
+def test_check_output_ref_annot():
 
     with tempfile.TemporaryDirectory() as wk:
 
@@ -52,6 +52,42 @@ def test_check_output():
              'strand_balance': '0.333'}
 
 
+def test_version():
+    cmd = "sequana_variant_calling --version"
+    subprocess.call(cmd.split())
+
+def test_check_output_no_annotation():
+
+    with tempfile.TemporaryDirectory() as wk:
+
+        cmd = "sequana_variant_calling --input-directory {} "
+        cmd += "--working-directory {} --run-mode local --force "
+        cmd += " --reference-file {}"
+        cmd = cmd.format(sharedir, wk, reference)
+        # create the wokring directory and script
+        subprocess.call(cmd.split())
+
+        subprocess.call("sh variant_calling.sh".split(), cwd=wk)
+
+        from sequana.freebayes_vcf_filter import VCF_freebayes, Variant
+        vcf = VCF_freebayes(wk + "/data/freebayes/data.raw.vcf")
+        vcf.rewind()
+        vv = [Variant(v)._resume for v in vcf]
+        # this may change depending on the freebayes version...
+        assert len(vv) in (67,)
+        vv = vv[0].copy()
+        del vv['freebayes_score']
+        assert vv == {'alternative': 'T',
+             'chr': 'JB409847',
+             'depth': 23,
+             #'freebayes_score': 2.78452e-14,
+             'type': 'SNV',
+             'frequency': '0.261',
+             'position': '2221',
+             'reference': 'C',
+             'strand_balance': '0.333'}
+
+
 def test_version():
     cmd = "sequana_variant_calling --version"
     subprocess.call(cmd.split())