Added globbing of input motifs, meaning that multiple motif file/dirs…

… are possible

Snakefile

@@ -16,7 +16,6 @@ CONFIGFILE = str(workflow.overwrite_configfile)
 
 #Snake modules used to setup run
 include: "snakefiles/helper.snake"
-
 #shell.prefix("")
 
 #-------------------------------------------------------------------------------#
@@ -54,7 +53,6 @@ else:
 	print("ERROR: Could not find any conditions (\"data:\{condition\}\") in configfile {0}".format(CONFIGFILE))
 	sys.exit()
 
-
 #-------------------------------------------------------------------------------#
 #------------------------- WHICH FILES/INFO WERE INPUT? ------------------------#
 #-------------------------------------------------------------------------------#
@@ -76,11 +74,10 @@ BLACKLIST = config['run_info']['blacklist']
 GTF = config['run_info']['gtf']
 OUTPUTDIR = config['run_info']["output"]
 BLACKLIST = config['run_info']['blacklist']
-MOTIFDIR = config['run_info']['motifs']
+#MOTIFS = config['run_info']['motifs']
 
 input_files.extend([FASTA, BLACKLIST, GTF])
 
-
 #---------- Test that input files exist -----------#
 for file in input_files:
 	if file != None:
@@ -89,12 +86,33 @@ for file in input_files:
 			exit("ERROR: The following file given in config does not exist: {0}".format(full_path))
 
 
-#--------------------------------- MOTIFS --------------------------------------#
+#--------------------------------- MOTIFS ------------------------------#
+
+#If not list, make it list and glob elements
+if not isinstance(config['run_info']['motifs'], list):
+	config['run_info']['motifs'] = [config['run_info']['motifs']]
+motif_input = sum([glob.glob(element) for element in config['run_info']['motifs']], [])
+
+#Test if input is directory or file
+motif_files = []
+for path in motif_input:
+
+	#If input is dir; fetch all input files
+	if os.path.isdir(path):
+		files = os.listdir(path)
+		motif_files.extend([os.path.join(path, f) for f in files])
+
+	#If input is file, add to list of files
+	elif os.path.isfile(path):
+		motif_files.append(path)
+
+motif_files = list(set(motif_files)) #remove duplicates
+config['run_info']['motifs'] = sorted(motif_files)
+
 #Identify IDS of motifs
-files = os.listdir(MOTIFDIR)
 MOTIF_FILES = {}
-for file in files:
-	full_file = os.path.join(MOTIFDIR, file)
+for file in motif_files:
+	full_file = file	
 	with open(full_file) as f:
 		for line in f:
 			if line.startswith("MOTIF"):
@@ -109,7 +127,6 @@ for file in files:
 
 TF_IDS = list(MOTIF_FILES.keys())
 
-
 #-------------------------------------------------------------------------------#
 #------------------------ WHICH FILES SHOULD BE CREATED? -----------------------#
 #-------------------------------------------------------------------------------#

example_config.yaml

@@ -7,12 +7,12 @@ data:
   Tcell: [data/Tcell_chr4_*.bam]  #list of .bam-files
 
 run_info:
-  organism: human                        #mouse/human
-  fasta: data/genome_chr4.fa.gz          #.fasta-file containing organism genome
-  blacklist: data/blacklist_chr4.bed     #.bed-file containing blacklisted regions
-  gtf: data/genes_chr4.gtf               #.gtf-file for annotation of peaks
-  motifs: data/individual_motifs         #directory containing motifs (single files in MEME/JASPAR/PFM format)
-  output: test_output                    #output directory 
+  organism: human                           #mouse/human
+  fasta: data/genome_chr4.fa.gz             #.fasta-file containing organism genome
+  blacklist: data/blacklist_chr4.bed        #.bed-file containing blacklisted regions
+  gtf: data/genes_chr4.gtf                  #.gtf-file for annotation of peaks
+  motifs: data/individual_motifs/*          #motifs (directory with files or individual files in MEME/JASPAR/PFM format)
+  output: test_output                       #output directory 
 
 
 #-------------------------------------------------------------------------#

snakefiles/footprinting.snake

@@ -5,7 +5,7 @@
 #Format motifs to pfm format
 rule format_motifs:
 	input: 
-		MOTIF_FILES.values()	#MOTIF_FILES is a dict
+		MOTIF_FILES.values()	#MOTIF_FILES is a dict with paths to motif files as values
 	output:
 		os.path.join(OUTPUTDIR, "motifs", "all_motifs.txt") 
 	priority: 2