Mercurial > repos > yhoogstrate > edger_with_design_matrix

annotate edger_dge_table_to_bedgraph @ 92:86c3aaa205b5 draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Added bedgraph exporting function for the DGE tables
author yhoogstrate
date Fri, 13 Mar 2015 05:45:09 -0400
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
92
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
1 #!/usr/bin/env python
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
2
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
3 import re,sys,os,os.path,argparse,textwrap,datetime
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
4
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
5 class GTF:
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
6 def __init__(self,filename,features=["exon"],symbol="gene_id"):
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
7 self.features = features
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
8 if(re.match("^[a-zA-Z0-9_\-]+$",symbol)):
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
9 self.symbol = symbol
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
10 else:
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
11 raise ValueError('False symbol matching symbol: '+str(symbol))
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
12 self.index = {}
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
13 self.parse(filename)
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
14
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
15 def parse(self,filename):
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
16 with open(filename) as infile:
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
17 for line in infile:
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
18 self.parse_line(line)
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
19
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
20 def parse_line(self,line):
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
21 line = line.strip().split("\t")
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
22 if(len(line) == 9):
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
23 if(line[2] in self.features):
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
24 gene_id = self.parse_column_9(line[8])
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
25 if(gene_id):
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
26 if(not self.index.has_key(gene_id)):
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
27 self.index[gene_id] = []
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
28 self.index[gene_id].append([line[0],line[3],line[4]])
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
29
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
30 def parse_column_9(self,line):
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
31 query = self.symbol+'[ =]+([^ ;]+)'
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
32 m = re.search(query, line)
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
33
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
34 if(m):
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
35 return m.group(1).strip("'").strip('"')
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
36 else:
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
37 return None
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
38
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
39 def get(self,key):
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
40 try:
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
41 return self.index[key]
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
42 except:
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
43 return False
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
44
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
45
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
46 class EdgeR_table:
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
47 def __init__(self,table,gtf,columns=[3,7]):
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
48 self.index = {}
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
49 self.gtf = gtf
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
50 self.columns = columns
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
51 self.parse(table)
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
52
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
53 def parse(self,filename):
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
54 i = 0
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
55 with open(filename) as infile:
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
56 for line in infile:
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
57 if(i == 0):
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
58 self.parse_header(line)
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
59 else:
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
60 self.parse_line(line)
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
61 i += 1
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
62
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
63 def parse_header(self,line):
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
64 params = line.strip().split("\t")
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
65 if(params[1].lower().find("genes") == -1):
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
66 raise ValueError('False header in file - no "genes" in 2nd colum: '+line)
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
67 if(params[2].lower().find("logfc") == -1):
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
68 raise ValueError('False header in file - no "logfc" in 3rd colum: '+line)
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
69 if(params[3].lower().find("logcpm") == -1):
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
70 raise ValueError('False header in file - no "logcpm" in 4th colum: '+line)
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
71 if(params[4].lower().find("lr") == -1):
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
72 raise ValueError('False header in file - no "lr" in 5th colum: '+line)
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
73 if(params[5].lower().find("pvalue") == -1):
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
74 raise ValueError('False header in file - no "pvalue" in 6th colum: '+line)
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
75 if(params[6].lower().find("fdr") == -1):
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
76 raise ValueError('False header in file - no "fdr" in 7th colum: '+line)
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
77
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
78 def parse_line(self,line):
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
79 line = line.strip().split("\t")
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
80
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
81 if(len(line) == 7):
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
82 gene_id = line[1].strip('"').strip("'")
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
83 column_data = {}
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
84 for column in self.columns:
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
85 if(column in [6,7]):
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
86 column_data[column] = str(1.0 - float(line[column-1]))
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
87 else:
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
88 column_data[column] = line[column-1]
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
89
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
90 locations = self.gtf.get(gene_id)
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
91 if(not locations):
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
92 print "Warning: no location found for gene "+gene_id
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
93 else:
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
94 for location in locations:
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
95 self.insert(location,column_data)
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
96
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
97 def insert(self,location,data):
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
98 chrom = location[0]
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
99 start = location[1]
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
100 end = location[2]
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
101
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
102 if(not self.index.has_key(chrom)):
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
103 self.index[chrom] = {}
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
104
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
105 if(not self.index[chrom].has_key(start)):
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
106 self.index[chrom][start] = {}
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
107
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
108 if(not self.index[chrom][start].has_key(end)):
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
109 self.index[chrom][start][end] = []
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
110
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
111 self.index[chrom][start][end].append(data)
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
112
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
113 def export(self,filenames={3:"log_cpm.txt",7:"fdr.txt"}):
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
114 for column in self.columns:
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
115 fh = open(filenames[column],"w")
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
116
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
117 buf = False
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
118
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
119 for chrom in sorted(self.index.keys()):
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
120 for start in sorted(self.index[chrom].keys()):
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
121 for end in sorted(self.index[chrom][start].keys()):
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
122 fh.write(chrom+"\t"+start+"\t"+end+"\t"+self.index[chrom][start][end][0][column]+"\n")
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
123
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
124 fh.close()
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
125
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
126 os.system("sort -k1,1V -k2,2g -k3,3g '"+filenames[column]+"' > '"+filenames[column]+".sorted'")
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
127
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
128
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
129 def remove_overlap_in_bedgraph(bedgraph_file_dirty,bedgraph_file_clean):
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
130 fh = open(bedgraph_file_clean,"w")
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
131 buf = False
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
132
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
133 with open(bedgraph_file_dirty,"r") as f:
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
134 for line in f:
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
135 cur = line.strip().split("\t")
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
136 cur[1] = int(cur[1])
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
137 cur[2] = int(cur[2])
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
138
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
139 if(not buf):
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
140 buf = cur
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
141 else:
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
142 if(cur[0] == buf[0] and cur[1] <= buf[2] ):
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
143 if(buf[1] == cur[1]): #is subset
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
144 newscore = (float(buf[3])+float(cur[3]))/2
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
145 buf[2] = cur[2]
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
146 buf[3] = newscore
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
147 else:
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
148 c1 = buf[1]
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
149 c2 = cur[1]
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
150 c3 = min(buf[2],cur[2])
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
151 c4 = max(buf[2],cur[2])
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
152
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
153 fh.write(buf[0]+"\t"+str(c1)+"\t"+str(c2-1)+"\t"+str(buf[3])+"\n")
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
154
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
155 newscore = (float(buf[3])+float(cur[3]))/2
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
156 #fh.write(buf[0]+"\t"+str(c2+1)+"\t"+str(c3)+"\t"+str(newscore)+"\tp2\n")
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
157 #buf = [buf[0], c3+1 , c4 , cur[3]]
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
158
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
159 buf = [buf[0], c2 , c4 , cur[3]]
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
160
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
161 # find if buf is a subset -> if so, merge and send to buffer
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
162 # or find the overlapping region
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
163
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
164 # if current is overlapping with buffer; merge:
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
165 ## [ ] < buf
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
166 ## [ ] < cur
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
167 ##
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
168 ## [ ] < buf
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
169 ## [ ] < cur
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
170 ## 111112222333333 << write 1 and 2 and keep 3 in buf
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
171 else:
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
172 fh.write(buf[0]+"\t"+str(buf[1])+"\t"+str(buf[2])+"\t"+str(buf[3])+"\n")
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
173 buf=cur
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
174
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
175 fh.write(buf[0]+"\t"+str(buf[1])+"\t"+str(buf[2])+"\t"+str(buf[3])+"\n")
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
176 fh.close()
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
177
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
178
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
179
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
180 if __name__ == "__main__":
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
181 parser = argparse.ArgumentParser()
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
182
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
183 parser = argparse.ArgumentParser(formatter_class=argparse.RawDescriptionHelpFormatter,epilog="For more info please visit:\n<https://github.com/yhoogstrate/fuma>")
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
184
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
185 parser.add_argument("-t",help="CPM table to extract columns from",nargs=1,required=True)
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
186 parser.add_argument("-g",help="GTF file used to extract genomic location",nargs=1,required=True)
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
187
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
188 parser.add_argument("-c3",help="Output (bedgraph) for column 3 (logFC)",nargs="?",required=False)
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
189 parser.add_argument("-c4",help="Output (bedgraph) for column 4 (logCPM)",nargs="?",required=False)
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
190 parser.add_argument("-c5",help="Output (bedgraph) for column 5 (LR)",nargs="?",required=False)
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
191 parser.add_argument("-c6",help="Output (bedgraph) for column 6 (PValue)",nargs="?",required=False)
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
192 parser.add_argument("-c7",help="Output (bedgraph) for column 7 (FDR)",nargs="?",required=False)
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
193
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
194 args = parser.parse_args()
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
195
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
196 #files = {3:"VCAP_logFC.hg19.bedgraph",7:"VCAP_fdr.hg19.bedgraph"}
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
197 files = {}
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
198
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
199 if(args.c3):
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
200 files[3] = args.c3
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
201 if(args.c4):
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
202 files[4] = args.c4
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
203 if(args.c5):
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
204 files[5] = args.c5
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
205 if(args.c6):
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
206 files[6] = args.c6
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
207 if(args.c7):
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
208 files[7] = args.c7
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
209
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
210 print "Parsing GTF file"
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
211 g = GTF(args.g[0])
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
212
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
213 print "Parsing EdgeR table"
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
214 e = EdgeR_table(args.t[0],g,files.keys())
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
215
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
216 print "Exporting raw bedgraph(s)"
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
217 e.export(files)
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
218
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
219 print "Removing overlapping entries in bedgraph(s)"
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
220 for key in files.keys():
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
221 remove_overlap_in_bedgraph(files[key]+".sorted",files[key])
86c3aaa205b5 Added bedgraph exporting function for the DGE tables
yhoogstrate
parents:
diff changeset
222 os.system("rm '"+files[key]+".sorted'")