Switch from v0.3 to latest git version of C++ code, requires a workaround to...

Switch from v0.3 to latest git version of C++ code, requires a workaround to fix the bps column of segmentation output.

Snake.config.json

@@ -2,8 +2,8 @@
     "chromosomes"   : ["chr1","chr2","chr3","chr4","chr5","chr6","chr7","chr8","chr9","chr10","chr11","chr12","chr13","chr14","chr15","chr16","chr17","chr18","chr19","chr20","chr21","chr22","chrX"],
 
     "reference"     : "/MMCI/TM/scratch/ref/GRCh38_no_alt_analysis_set/GCA_000001405.15_GRCh38_no_alt_analysis_set.fna",
-    "mosaicatcher"  : "/MMCI/TM/scratch/strandseq/mosaicatcher-builds/v0.3/build/mosaic",
-    "plot_script"   : "/MMCI/TM/scratch/strandseq/mosaicatcher-builds/v0.3/R/qc.R",
+    "mosaicatcher"  : "/MMCI/TM/scratch/strandseq/mosaicatcher-builds/git/build/mosaic",
+    "plot_script"   : "/MMCI/TM/scratch/strandseq/mosaicatcher-builds/git/R/qc.R",
     "samtools"      : "samtools",
     "bcftools"      : "bcftools",
 

Snakefile

@@ -490,7 +490,7 @@ rule segmentation:
     input:
         "counts/{sample}/{window}_{file_name}.txt.gz"
     output:
-        "segmentation/{sample}/{window,\d+}_{file_name}.txt"
+        "segmentation/{sample}/{window,\d+}_{file_name}.txt.fixme"
     log:
         "log/segmentation/{sample}/{window}_{file_name}.log"
     params:
@@ -506,6 +506,16 @@ rule segmentation:
         {input} > {log} 2>&1
         """
 
+# TODO: This is a workaround because latest versions of "mosaic segment" don't compute the "bps"
+# TODO: column properly. Remove once fixed in the C++ code.
+rule fix_segmentation:
+    input:
+        "segmentation/{sample}/{window}_{file_name}.txt.fixme"
+    output:
+        "segmentation/{sample}/{window,\d+}_{file_name}.txt"
+    shell:
+        'awk \'BEGIN {{OFS="\\t"}} {{if ($1=="{wildcards.sample}") $12=int(($14-1)/100000); print}}\' {input} > {output}'
+
 # Pick a few segmentations and prepare the input files for SV classification
 rule prepare_segments:
     input:
@@ -523,7 +533,7 @@ rule segment_one_cell:
     input:
         "counts-per-cell/{sample}/{cell}/{window}_{file_name}.txt.gz"
     output:
-        "segmentation-per-cell/{sample}/{cell}/{window,\d+}_{file_name}.txt"
+        "segmentation-per-cell/{sample}/{cell}/{window,\d+}_{file_name}.txt.fixme"
     log:
         "log/segmentation-per-cell/{sample}/{cell}/{window}_{file_name}.log"
     params:
@@ -539,6 +549,16 @@ rule segment_one_cell:
         {input} > {log} 2>&1
         """
 
+# TODO: This is a workaround because latest versions of "mosaic segment" don't compute the "bps"
+# TODO: column properly. Remove once fixed in the C++ code.
+rule fix_segmentation_one_cell:
+    input:
+        "segmentation-per-cell/{sample}/{cell}/{window}_{file_name}.txt.fixme"
+    output:
+        "segmentation-per-cell/{sample}/{cell}/{window,\d+}_{file_name}.txt"
+    shell:
+        'awk \'BEGIN {{OFS="\\t"}} {{if ($1=="{wildcards.sample}") $12=int(($14-1)/100000); print}}\' {input} > {output}'
+
 rule segmentation_selection:
     input:
         counts="counts/{sample}/{window}_{file_name}.txt.gz",