|
5
|
1 from viz_graphs import *
|
|
|
2 from viz_functions import *
|
|
|
3 import argparse
|
|
|
4 import time
|
|
|
5 from multiprocessing import Process, Lock, Manager
|
|
|
6
|
|
|
7 ##################################################################################################################################################################################################################################################################
|
|
|
8 starttime = time.time()
|
|
|
9
|
|
|
10 parser = argparse.ArgumentParser()
|
|
12
|
11 parser.add_argument("-in", "--input", help="input file from Deseq2 or EdgeR", action="store")
|
|
|
12 parser.add_argument("-p_value", "--pval", help="p-value or p-adj threshold", action="store")
|
|
|
13 parser.add_argument("-fc", "--log2fc", help="log2fc threshold", action="store")
|
|
|
14 parser.add_argument("-top", "--top_mirnas", help="Top significant genes", action="store")
|
|
5
|
15 parser.add_argument("-tool_dir", "--tool_directory", help="tool directory path", action="store")
|
|
12
|
16 parser.add_argument("-statistic", "--stat", help="Filter option by p-value or p-adj or basemean (only for Deseq2)", action="store")
|
|
|
17 parser.add_argument("-diff_tool", "--tool", help="Choose tool for Differential expression Deseq2 or EdgeR", action="store")
|
|
5
|
18
|
|
|
19 args = parser.parse_args()
|
|
|
20
|
|
|
21 l=Lock()
|
|
|
22 number = int(args.top_mirnas)
|
|
|
23 log2fc = float(args.log2fc)
|
|
|
24 pval = float(args.pval)
|
|
|
25 iso_star_flag=0
|
|
|
26 non_star_flag=0
|
|
|
27
|
|
|
28 if args.tool=="2":
|
|
|
29
|
|
|
30 raw_EdgeR = read(args.input,0)
|
|
|
31 EdgeR = [x.rstrip("\n").split("\t") for x in raw_EdgeR]
|
|
|
32 del EdgeR[0]
|
|
|
33 for x in EdgeR:
|
|
|
34 if "/" in x[0]:
|
|
|
35 x[0]=x[0].split("/")[0]+"★"
|
|
|
36
|
|
|
37 if args.stat=="1":
|
|
|
38 non_templated = [[x[0],x[1],x[4]] for x in EdgeR if "nt" in x[0] and x[1]!="NA" and x[4]!="NA"]
|
|
|
39 matures = [[x[0],x[1],x[4]] for x in EdgeR if 'chr' in x[0].split("_")[-1] and "nt" not in x[0] and x[1]!="NA" and x[4]!="NA"]
|
|
|
40 isoforms = [[x[0],x[1],x[4]] for x in EdgeR if 'chr' not in x[0].split("_")[-1] and "nt" not in x[0] and x[1]!="NA" and x[4]!="NA"]
|
|
|
41 else:
|
|
|
42 non_templated = [[x[0],x[1],x[5]] for x in EdgeR if "nt" in x[0] and x[1]!="NA" and x[5]!="NA"]
|
|
|
43 matures = [[x[0],x[1],x[5]] for x in EdgeR if 'chr' in x[0].split("_")[-1] and "nt" not in x[0] and x[1]!="NA" and x[5]!="NA"]
|
|
|
44 isoforms = [[x[0],x[1],x[5]] for x in EdgeR if 'chr' not in x[0].split("_")[-1] and "nt" not in x[0] and x[1]!="NA" and x[5]!="NA"]
|
|
|
45
|
|
|
46 if args.tool=="1":
|
|
|
47
|
|
|
48 raw_Deseq = read(args.input,0)
|
|
|
49 Deseq = [x.rstrip("\n").split("\t") for x in raw_Deseq]
|
|
|
50 for x in Deseq:
|
|
|
51 if "/" in x[0]:
|
|
|
52 x[0]=x[0].split("/")[0]+"★"
|
|
|
53
|
|
|
54 if args.stat=="1":
|
|
|
55 non_templated = [[x[0],x[2],x[5]] for x in Deseq if "nt" in x[0] and x[2]!="NA" and x[5]!="NA"]
|
|
|
56 matures = [[x[0],x[2],x[5]] for x in Deseq if 'chr' in x[0].split("_")[-1] and "nt" not in x[0] and x[2]!="NA" and x[5]!="NA"]
|
|
|
57 isoforms = [[x[0],x[2],x[5]] for x in Deseq if 'chr' not in x[0].split("_")[-1] and "nt" not in x[0] and x[2]!="NA" and x[5]!="NA"]
|
|
|
58 elif args.stat=="2":
|
|
|
59 non_templated = [[x[0],x[2],x[6]] for x in Deseq if "nt" in x[0] and x[2]!="NA" and x[6]!="NA"]
|
|
|
60 matures = [[x[0],x[2],x[6]] for x in Deseq if 'chr' in x[0].split("_")[-1] and "nt" not in x[0] and x[2]!="NA" and x[6]!="NA"]
|
|
|
61 isoforms = [[x[0],x[2],x[6]] for x in Deseq if 'chr' not in x[0].split("_")[-1] and "nt" not in x[0] and x[2]!="NA" and x[6]!="NA"]
|
|
|
62 else:
|
|
|
63 non_templated = [[x[0],x[2],x[1]] for x in Deseq if "nt" in x[0] and x[2]!="NA" and x[1]!="NA"]
|
|
|
64 matures = [[x[0],x[2],x[1]] for x in Deseq if 'chr' in x[0].split("_")[-1] and "nt" not in x[0] and x[2]!="NA" and x[1]!="NA"]
|
|
|
65 isoforms = [[x[0],x[2],x[1]] for x in Deseq if 'chr' not in x[0].split("_")[-1] and "nt" not in x[0] and x[2]!="NA" and x[1]!="NA"]
|
|
|
66
|
|
|
67
|
|
|
68 diff_matures,diff_isoforms,diff_non_templated,names,non_temp,mat_iso = preproccess(non_templated,matures,isoforms,log2fc,pval,args.stat)
|
|
|
69 for x in mat_iso[:number]:
|
|
|
70 if "★" in x[0]:
|
|
|
71 iso_star_flag=1
|
|
|
72 break
|
|
|
73 for x in non_temp[:number]:
|
|
|
74 if "★" in x[0]:
|
|
|
75 non_star_flag=1
|
|
|
76 break
|
|
|
77
|
|
|
78
|
|
|
79 if non_templated!=[]:
|
|
|
80 analysis="2"
|
|
|
81 p=[Process(target=top_diff,args=(non_temp,number,"nt",l))]
|
|
|
82 p.extend([Process(target=top_diff,args=(mat_iso,number,"t",l))])
|
|
|
83 p.extend([Process(target=top_scatter_non,args=(diff_matures,diff_isoforms,diff_non_templated,names,number))])
|
|
|
84
|
|
|
85 else:
|
|
|
86 analysis="1"
|
|
|
87 p=[Process(target=top_diff,args=(mat_iso,number,"t",l))]
|
|
|
88 p.extend([Process(target=top_scatter_tem,args=(diff_matures,diff_isoforms,names,number))])
|
|
|
89
|
|
|
90 [x.start() for x in p]
|
|
|
91 [x.join() for x in p]
|
|
|
92
|
|
|
93 pdf_after_DE(analysis,args.top_mirnas,args.tool_directory,iso_star_flag,non_star_flag)
|
|
|
94
|
|
11
|
95 print('Runtime: {} seconds'.format(round(time.time() - starttime,2)))
|
|
5
|
96
|
