Fixed incorrect sorting of bound/unbound; Removed heatmap comparison …

…within conditions

snakemake_pipeline/TOBIAS.snake

@@ -113,7 +113,7 @@ for file in files:
 TF_IDS = list(MOTIF_FILES.keys())
 
 
-#---------------------------- OUTPUT PER CONDITION -----------------------------#
+#----------------------------------- OUTPUT  -----------------------------------#
 
 id2bam = {condition:{} for condition in CONDITION_IDS}
 
@@ -124,13 +124,12 @@ for condition in CONDITION_IDS:
 	id2bam[condition] = {sampleids[i]:config_bams[i] for i in range(len(sampleids))}	# Link sample ids to bams
 
 
-PLOTNAMES = expand("{condition}_{plotname}", condition=CONDITION_IDS, plotname=["heatmap", "aggregate"])
+PLOTNAMES = expand("{condition}_{plotname}", condition=CONDITION_IDS, plotname=["aggregate"])
 if len(CONDITION_IDS) > 1:
-	PLOTNAMES.extend(["heatmap_comparison", "aggregate_comparison"]) 
+	PLOTNAMES.extend(["heatmap_comparison", "aggregate_comparison_all", "aggregate_comparison_bound"]) 
 
 output_files.append(expand(os.path.join(OUTPUTDIR, "footprinting", "{condition}_footprints.bw"), condition=CONDITION_IDS))
 
-#output_files.append(os.path.join(OUTPUTDIR, "overview", "TFBS_distance.txt"))
 output_files.append(os.path.join(OUTPUTDIR, "TFBS", "bindetect_results.txt"))
 output_files.append(os.path.join(OUTPUTDIR, "overview", "bindetect_results.txt"))
 
@@ -139,21 +138,6 @@ output_files.extend(expand(os.path.join(OUTPUTDIR, "TFBS", "{TF}", "plots", "{TF
 output_files.extend(expand(os.path.join(OUTPUTDIR, "overview", "all_{plotname}.pdf"), plotname=PLOTNAMES))
 
 
-#-------------------------- OUTPUT ACROSS CONDITIONS ---------------------------#
-
-"""
-COMPARE_COND = 0
-if len(CONDITION_IDS) > 1:
-	COMPARE_COND = 1 	# flag
-	output_files.extend(expand(os.path.join(OUTPUTDIR, "TFBS", "{TF}", "plots", "{TF}_heatmap_comparison.pdf"), TF=TF_IDS))
-	output_files.extend(expand(os.path.join(OUTPUTDIR, "TFBS", "{TF}", "plots", "{TF}_aggregate_comparison.pdf"), TF=TF_IDS))
-	#output_files.extend([os.path.join(OUTPUTDIR, "overview", "diff_bind_plot.pdf")])
-
-"""
-#-------------------------------- OTHER OUTPUT ---------------------------------#
-
-
-
 
 #-------------------------------------------------------------------------------#
 #--------------------- WHICH SNAKE MODULES SHOULD BE USED? ---------------------#
@@ -164,12 +148,11 @@ include: "snakefiles/footprinting.snake"
 include: "snakefiles/visualization.snake"
 
 
-
 #-------------------------------------------------------------------------------#
 #------------------------ DEAL WITH SPECIAL ENVIRONMENTS -----------------------#
 #-------------------------------------------------------------------------------#
 
-"""
+
 sys_env = subprocess.check_output(['conda', 'env', 'list'], universal_newlines=True)
 env_list = [line.split()[0] for line in sys_env.split("\n") if len(line.split()) > 0]
 
@@ -183,7 +166,7 @@ if "MACS_ENV" not in env_list:
 	print("Creating macs environment for the first time")
 	subprocess.call(["conda", "env", "create", "--file", "environments/macs.yaml"])
 
-"""
+
 #-------------------------------------------------------------------------------#
 #---------------------------------- RUN :-) ------------------------------------#
 #-------------------------------------------------------------------------------#

snakemake_pipeline/snakefiles/visualization.snake

@@ -1,7 +1,7 @@
 #----------------------------------------------------------------#
 #---------------------------- Heatmaps --------------------------#
 #----------------------------------------------------------------#
-
+"""
 #Heatmaps split in bound/unbound within conditions
 rule plot_heatmaps_within:
 	input:
@@ -16,7 +16,7 @@ rule plot_heatmaps_within:
 		"--title '{TF} heatmap'",
 	shell:
 		"TOBIAS PlotHeatmap --TFBS {input.beds} --signals {input.tracks} --output {output.heatmap} {params} >/dev/null"
-
+"""
 
 #Heatmaps across conditions
 rule plot_heatmaps_across:
@@ -44,8 +44,7 @@ rule plot_heatmaps_across:
 rule plot_aggregate_within:
 	input: 
 		TFBS = [os.path.join(OUTPUTDIR, "TFBS", "{TF}", "beds", "{TF}_all.bed")] + [os.path.join(OUTPUTDIR, "TFBS", "{TF}", "beds", "{TF}_{condition}_" + state + ".bed") for state in ["bound", "unbound"]],
-		signals = [os.path.join(OUTPUTDIR, "bias_correction", "{condition}_" + state + ".bw") for state in ["uncorrected", "bias", "expected", "corrected"]],
-		#whitelist = os.path.join(OUTPUTDIR, "peaks", condition + "_union.bed")
+		signals = [os.path.join(OUTPUTDIR, "bias_correction", "{condition}_" + state + ".bw") for state in ["uncorrected", "expected", "corrected"]],
 	output:
 		os.path.join(OUTPUTDIR, "TFBS", "{TF}", "plots", "{TF}_{condition}_aggregate.pdf")
 	message: "Plotting split between bound/unbound around TFBS for TF \"{wildcards.TF}\" in condition \"{wildcards.condition}\""
@@ -57,25 +56,24 @@ rule plot_aggregate_within:
 		"TOBIAS PlotAggregate --TFBS {input.TFBS} --signals {input.signals} --output {output} {params} >/dev/null "
 
 
-#Aggregates across conditions
+#Aggregates across conditions for all and for bound subsets
 rule plot_aggregate_across:
 	input: 
 		TFBS_all = os.path.join(OUTPUTDIR, "TFBS", "{TF}", "beds", "{TF}_all.bed"),
 		TFBS_bound = expand(os.path.join(OUTPUTDIR, "TFBS", "{{TF}}", "beds", "{{TF}}_{condition}_bound.bed"), condition=CONDITION_IDS),
 		signals = expand(os.path.join(OUTPUTDIR, "bias_correction", "{condition}_corrected.bw"), condition=CONDITION_IDS),
-		#regions = os.path.join(OUTPUTDIR, "peaks", condition + "_union.bed") #[os.path.join(OUTPUTDIR, "peaks", condition + "_union.bed") for condition in CONDITION_IDS]
 	output: 
-		all_compare = os.path.join(OUTPUTDIR, "TFBS", "{TF}", "plots", "{TF}_aggregate_comparison.pdf"),
-		#bound_compare = os.path.join(OUTPUTDIR, "TFBS", "{TF}", "plots", "{TF}_aggregate_comparison_bound.pdf")
+		all_compare = os.path.join(OUTPUTDIR, "TFBS", "{TF}", "plots", "{TF}_aggregate_comparison_all.pdf"),
+		bound_compare = os.path.join(OUTPUTDIR, "TFBS", "{TF}", "plots", "{TF}_aggregate_comparison_bound.pdf")
 	priority: 2
 	params: 
 		"--title {0}".format("{TF}"),
 		"--plot_boundaries",
 		"--share_y both",
 	message: "Plotting comparison of cutsite signals for \"{wildcards.TF}\" between conditions"
 	shell:
-		"TOBIAS PlotAggregate --TFBS {input.TFBS_all} {input.TFBS_bound} --signals {input.signals} --output {output.all_compare} {params} >/dev/null"
-		#"TOBIAS PlotAggregate --TFBS {input.TFBS_bound} --signals {input.signals} --output {output.bound_compare} {params};"
+		"TOBIAS PlotAggregate --TFBS {input.TFBS_all} --signals {input.signals} --output {output.all_compare} {params} >/dev/null; "
+		"TOBIAS PlotAggregate --TFBS {input.TFBS_bound} --signals {input.signals} --output {output.bound_compare} {params}  >/dev/null;"
 
 
 #----------------------------------------------------------------#