added pre batch removal tpms

Author: Ubuntu

modules/preprocess/preprocess_cohort.snakefile

@@ -26,8 +26,8 @@ def preprocess_cohort_targets(wildcards):
     ls.append("analysis/batchremoval/%s_%s_tpm.genesymbol.csv" % (design,covariates))
     ls.append("analysis/batchremoval/%s_%s_tpm.genesymbol.batchremoved.csv" % (design,covariates))
 
-    ls.append("analysis/batchremoval/pca_plot_before.png" )
-    ls.append("analysis/batchremoval/pca_plot_after.png" )
+    ls.append("analysis/batchremoval/%s_%s_pca_plot_before.png" % (design,covariates))
+    ls.append("analysis/batchremoval/%s_%s_pca_plot_after.png" % (design,covariates))
     return ls 
 
 rule preprocess_cohort_all:
@@ -154,22 +154,23 @@ rule batch_removal:
 
 rule pca_sample_clustering:
     input:
-        before_batch = "analysis/salmon/{design}_{covariates}_tpm.genesymbol.csv",
-        after_batch = "analysis/batchremoval/{design}_{covariates}_tpm.genesymbol.batchremoved.csv"
+        before_batch = "analysis/batchremoval/{design}_{covariates}_tpm.genesymbol.batchremoved.csv",
+        after_batch = "analysis/batchremoval/{design}_{covariates}_tpm.genesymbol.csv"
     output:
-        "analysis/batchremoval/{design}_{covariates}_pca_plot_before.png",
-        "analysis/batchremoval/{design}_{covariates}_pca_plot_after.png",
+        before_pca = "analysis/batchremoval/{design}_{covariates}_pca_plot_before.png",
+        after_pca = "analysis/batchremoval/{design}_{covariates}_pca_plot_after.png",
     message:
         "Running PCA for sample clustering"
     benchmark:
         "benchmarks/batchremoval/{design}_{covariates}_pca.benchmark"
     params:
         meta_info = config["metasheet"],
-        out_path = "analysis/batchremoval/",
         path="set +eu;source activate %s" % config['stat_root'],
         covariates = config["batch"],
+        design = config["design"]
     conda: "../envs/stat_perl_r.yml"
     shell:
-        "{params.path}; Rscript src/preprocess/pca.R -b {input.before_batch} -a {input.after_batch} -m {params.meta_info} -o {params.out_path} -c {params.covariates}"
+        "{params.path}; Rscript src/preprocess/pca.R -b {input.before_batch} -a {input.after_batch} -m {params.meta_info}  -c {params.covariates} \
+                -g {params.design} -i {output.before_pca} -j {output.after_pca}"
 
 

src/preprocess/pca.R

@@ -12,10 +12,14 @@ option_list = list(
               help="expression matrix after batch removal", metavar="character"),
   make_option(c("-m", "--metainfo"), type="character", default=NULL, 
               help="meta file", metavar="character"),
-  make_option(c("-o", "--outdir"), type="character", default=NULL, 
-              help="output directory", metavar="character"),
+  make_option(c("-i", "--out_before"), type="character", default=NULL, 
+              help="PCA plot before batch removal", metavar="character"),
+  make_option(c("-j", "--out_after"), type="character", default=NULL,
+              help="PCA plot after batch removal", metavar="character"),
   make_option(c("-c", "--column"), type="character", default=NULL, 
-              help="batch column", metavar="character")   
+              help="batch column", metavar="character"),
+  make_option(c("-g", "--group"), type="character", default=NULL,
+              help="group column", metavar="character")  
 ); 
 
 opt_parser = OptionParser(option_list=option_list);
@@ -24,7 +28,7 @@ opt = parse_args(opt_parser);
 expr_before <- read.table(opt$before,sep = ",", row.names = 1, header = TRUE, check.names = FALSE)
 expr_after <- read.table(opt$after,sep = ",", row.names = 1, header = TRUE, check.names = FALSE)
 annot <- read.table(opt$metainfo,sep = ",",header = TRUE, row.names = 1)
-pca_out_dir <- opt$outdir
+
 
 if (opt$column == "False") {
   annot$Batch <- 1
@@ -54,12 +58,12 @@ pca_plot <- function(exprTable, annot,title, Batch) {
 
 
 ###print pca plot
-png(file = paste(pca_out_dir,"pca_plot_before.png",sep = ""), res = 300, height = 1200, width = 1500)
+png(file = opt$out_before, res = 300, height = 1200, width = 1500)
 pca_plot(exprTable = expr_before, annot = annot, title = "PCA plot Before Batch Removal",Batch=Batch)
 dev.off()
 
-png(file = paste(pca_out_dir,"pca_plot_after.png",sep = ""), res = 300, height = 1200, width = 1500)
+png(file = opt$out_after, res = 300, height = 1200, width = 1500)
 pca_plot(exprTable = expr_after, annot = annot, title = "PCA plot After Batch Removal",Batch = Batch)
 dev.off()
 
-###
+###