[WIP] WES workflow

lib/common.py

@@ -677,6 +677,25 @@ def is_paired_end(sampletable, sample):
         pass
     return False
 
+def is_tumor_only(sampletable):
+    """
+    For somatic WES pipeline, inspects the sampletable to see if there are only
+    tumor files rather than paired tumor/normal. Assumes the presence of
+    a column named 'normal_filename' indicates that there are paired normal
+    samples for all tumor samples. Does not support mixing tumor/normal and
+    tumor only samples in the same sampletable.
+
+    Parameters
+    ----------
+    sampletable : pandas.DataFrame
+        Only contains columns called 'normal_filename' and (if paired end)
+        'normal_R2_filename' if there are paired normal samples for every tumor
+        sample. Does not support blank/NA 'normal_filename' columns. 
+    """
+    if 'normal_filename' in sampletable.columns:
+        return False
+    return True
+
 
 def fill_r1_r2(sampletable, pattern, r1_only=False):
     """

lib/patterns_targets.py

@@ -69,6 +69,35 @@ def __init__(self, config, patterns, workdir=None):
             self.n = [1]
 
 
+class WESConfig(SeqConfig):
+    def __init__(self, config, patterns, workdir=None):
+        """
+        Config object specific to WES workflows.
+
+        Fills in patterns to create targets
+
+        Parameters
+        ----------
+
+        config : dict
+
+        patterns : str
+            Path to patterns YAML file
+
+        workdir : str
+            Config, patterns, and all paths in `config` should be interpreted
+            as relative to `workdir`
+        """
+        SeqConfig.__init__(self, config, patterns, workdir)
+        self.tumoronly = common.is_tumor_only(self.sampletable)
+        if self.tumoronly:
+            self.genotype = ['tumor']
+        else:
+            self.genotype = ['tumor', 'normal']
+        self.fill = dict(sample=self.samples, genotype=self.genotype, n=self.n)
+        self.targets = helpers.fill_patterns(self.patterns, self.fill, zip)
+
+
 class RNASeqConfig(SeqConfig):
     def __init__(self, config, patterns, workdir=None):
         """

requirements.txt

@@ -1,4 +1,5 @@
 atropos
+bcftools
 bedtools
 bioconductor-annotationhub
 bioconductor-biocparallel
@@ -14,6 +15,7 @@ bioconductor-sva
 bioconductor-tximport
 biopython >=1.68
 bowtie2
+bwa
 cutadapt
 deeptools >=3.0.1
 fastqc
@@ -23,6 +25,7 @@ fastq-screen
 font-ttf-dejavu-sans-mono
 
 gat
+gatk4
 gffutils >=0.8.7.1
 ghostscript
 git
@@ -69,6 +72,8 @@ samtools >=1.4.1
 scipy >=0.18.1
 seaborn >=0.7.1
 snakemake==5.5.4
+snpeff
+somatic-sniper
 sra-tools
 star
 subread
@@ -84,3 +89,5 @@ ucsc-liftover
 ucsc-oligomatch
 ucsc-twobittofa
 ucsc-wigtobigwig
+varscan
+vcftools