view bin/edger_dge_table_to_bedgraph @ 5:bde663b872d9 draft

planemo upload for repository https://bitbucket.org/EMCbioinf/galaxy-tool-shed-tools/raw/master/edger_with_design_matrix commit 275a72ec0424e4e5d658d1bc8227077ea46f0fdc

#!/usr/bin/env python

import re,sys,os,os.path,argparse,textwrap,datetime

class GTF:
	def __init__(self,filename,features=["exon"],symbol="gene_id"):
		self.features = features
		if(re.match("^[a-zA-Z0-9_\-]+$",symbol)):
			self.symbol = symbol
		else:
			raise ValueError('False symbol matching symbol: '+str(symbol))
		self.index = {}
		self.parse(filename)
	
	def parse(self,filename):
		with open(filename) as infile:
			for line in infile:
				self.parse_line(line)
	
	def parse_line(self,line):
		line = line.strip().split("\t")
		if(len(line) == 9):
			if(line[2] in self.features):
				gene_id = self.parse_column_9(line[8])
				if(gene_id):
					if(not self.index.has_key(gene_id)):
						self.index[gene_id] = []
					self.index[gene_id].append([line[0],line[3],line[4]])
	
	def parse_column_9(self,line):
		query = self.symbol+'[ =]+([^ ;]+)'
		m = re.search(query, line)
		
		if(m):
			return m.group(1).strip("'").strip('"')
		else:
			return None
	
	def get(self,key):
		try:
			return self.index[key]
		except:
			return False


class EdgeR_table:
	def __init__(self,table,gtf,columns=[3,7]):
		self.index = {}
		self.gtf = gtf
		self.columns = columns
		self.parse(table)
	
	def parse(self,filename):
		i = 0
		with open(filename) as infile:
			for line in infile:
				if(i == 0):
					self.parse_header(line)
				else:
					self.parse_line(line)
				i += 1
	
	def parse_header(self,line):
		params = line.strip().split("\t")
		if(params[1].lower().find("genes") == -1):
			raise ValueError('False header in file - no "genes" in 2nd colum: '+line)
		if(params[2].lower().find("logfc") == -1):
			raise ValueError('False header in file - no "logfc" in 3rd colum: '+line)
		if(params[3].lower().find("logcpm") == -1):
			raise ValueError('False header in file - no "logcpm" in 4th colum: '+line)
		if(params[4].lower().find("lr") == -1):
			raise ValueError('False header in file - no "lr" in 5th colum: '+line)
		if(params[5].lower().find("pvalue") == -1):
			raise ValueError('False header in file - no "pvalue" in 6th colum: '+line)
		if(params[6].lower().find("fdr") == -1):
			raise ValueError('False header in file - no "fdr" in 7th colum: '+line)

	def parse_line(self,line):
		line = line.strip().split("\t")
		
		if(len(line) == 7):
			gene_id = line[1].strip('"').strip("'")
			column_data = {}
			for column in self.columns:
				if(column in [6,7]):
					column_data[column] = str(1.0 - float(line[column-1]))
				else:
					column_data[column] = line[column-1]
			
			locations = self.gtf.get(gene_id)
			if(not locations):
				print "Warning: no location found for gene "+gene_id
			else:
				for location in locations:
					self.insert(location,column_data)
	
	def insert(self,location,data):
		chrom = location[0]
		start = location[1]
		end = location[2]
		
		if(not self.index.has_key(chrom)):
			self.index[chrom] = {}
		
		if(not self.index[chrom].has_key(start)):
			self.index[chrom][start] = {}
		
		if(not self.index[chrom][start].has_key(end)):
			self.index[chrom][start][end] = []
		
		self.index[chrom][start][end].append(data)
	
	def export(self,filenames={3:"log_cpm.txt",7:"fdr.txt"}):
		for column in self.columns:
			fh = open(filenames[column],"w")
			
			buf = False
			
			for chrom in sorted(self.index.keys()):
				for start in sorted(self.index[chrom].keys()):
					for end in sorted(self.index[chrom][start].keys()):
						fh.write(chrom+"\t"+start+"\t"+end+"\t"+self.index[chrom][start][end][0][column]+"\n")
			
			fh.close()
			
			os.system("sort -k1,1V -k2,2g -k3,3g '"+filenames[column]+"' > '"+filenames[column]+".sorted'")


def remove_overlap_in_bedgraph(bedgraph_file_dirty,bedgraph_file_clean):
	fh = open(bedgraph_file_clean,"w")
	buf = False
	
	with open(bedgraph_file_dirty,"r") as f:
		for line in f:
			cur = line.strip().split("\t")
			cur[1] = int(cur[1])
			cur[2] = int(cur[2])
			
			if(not buf):
				buf = cur
			else:
				if(cur[0] == buf[0] and cur[1] <= buf[2] ):
					if(buf[1] == cur[1]):									#is subset
						newscore = (float(buf[3])+float(cur[3]))/2
						buf[2] = cur[2]
						buf[3] = newscore
					else:
						c1 = buf[1]
						c2 = cur[1]
						c3 = min(buf[2],cur[2])
						c4 = max(buf[2],cur[2])
						
						fh.write(buf[0]+"\t"+str(c1)+"\t"+str(c2-1)+"\t"+str(buf[3])+"\n")
						
						newscore = (float(buf[3])+float(cur[3]))/2
						#fh.write(buf[0]+"\t"+str(c2+1)+"\t"+str(c3)+"\t"+str(newscore)+"\tp2\n")
						#buf = [buf[0], c3+1 , c4 , cur[3]]
						
						buf = [buf[0], c2 , c4 , cur[3]]
						
						# find if buf is a subset -> if so, merge and send to buffer
						# or find the overlapping region
						
						# if current is overlapping with buffer; merge:
						##  [       ]                 < buf
						##       [        ]           < cur
						##
						##  [             ]           < buf
						##       [    ]               < cur
						##  111112222333333  << write 1 and 2 and keep 3 in buf
				else:
					fh.write(buf[0]+"\t"+str(buf[1])+"\t"+str(buf[2])+"\t"+str(buf[3])+"\n")
					buf=cur
		
		fh.write(buf[0]+"\t"+str(buf[1])+"\t"+str(buf[2])+"\t"+str(buf[3])+"\n")
	fh.close()



if __name__ == "__main__":
	parser = argparse.ArgumentParser()
	
	parser = argparse.ArgumentParser(formatter_class=argparse.RawDescriptionHelpFormatter,epilog="For more info please visit:\n<https://github.com/yhoogstrate/fuma>")
	
	parser.add_argument("-t",help="CPM table to extract columns from",nargs=1,required=True)
	parser.add_argument("-g",help="GTF file used to extract genomic location",nargs=1,required=True)
	
	parser.add_argument("-c3",help="Output (bedgraph) for column 3 (logFC)",nargs="?",required=False)
	parser.add_argument("-c4",help="Output (bedgraph) for column 4 (logCPM)",nargs="?",required=False)
	parser.add_argument("-c5",help="Output (bedgraph) for column 5 (LR)",nargs="?",required=False)
	parser.add_argument("-c6",help="Output (bedgraph) for column 6 (PValue)",nargs="?",required=False)
	parser.add_argument("-c7",help="Output (bedgraph) for column 7 (FDR)",nargs="?",required=False)
	
	args = parser.parse_args()
	
	#files = {3:"VCAP_logFC.hg19.bedgraph",7:"VCAP_fdr.hg19.bedgraph"}
	files = {}
	
	if(args.c3):
		files[3] = args.c3
	if(args.c4):
		files[4] = args.c4
	if(args.c5):
		files[5] = args.c5
	if(args.c6):
		files[6] = args.c6
	if(args.c7):
		files[7] = args.c7
	
	print "Parsing GTF file"
	g = GTF(args.g[0])
	
	print "Parsing EdgeR table"
	e = EdgeR_table(args.t[0],g,files.keys())
	
	print "Exporting raw bedgraph(s)"
	e.export(files)
	
	print "Removing overlapping entries in bedgraph(s)"
	for key in files.keys():
		remove_overlap_in_bedgraph(files[key]+".sorted",files[key])
		os.system("rm '"+files[key]+".sorted'")