|
2
|
1
|
|
|
2 import os,sys
|
|
|
3
|
|
|
4 file1 = sys.argv[1]
|
|
|
5 file2 = sys.argv[2]
|
|
|
6 genome = sys.argv[3]
|
|
|
7 outplot = sys.argv[4]
|
|
|
8 outlog = sys.argv[5]
|
|
|
9 outbed = sys.argv[6]
|
|
|
10
|
|
|
11 strandness = ''
|
|
|
12 if len(sys.argv) > 7:
|
|
|
13 strandness = sys.argv[7]
|
|
|
14
|
|
|
15 # real distance
|
|
|
16 cmd = 'closestBed -a '+file1+' -b '+file2 + ' '+strandness + ' -d -t first > '+outbed
|
|
|
17 os.system(cmd)
|
|
|
18 # shuffle
|
|
|
19 cmd = 'shuffleBed -chrom -g '+genome+' -i '+file1+'> shuffled.bed'
|
|
|
20 os.system(cmd)
|
|
|
21 # shuffled distance
|
|
|
22 cmd = 'closestBed -a shuffled.bed -b '+file2 + ' '+strandness + ' -d -t first > shuffled.dist'
|
|
|
23 os.system(cmd)
|
|
|
24
|
|
|
25
|
|
|
26 # test in R
|
|
|
27 r = open('tmp.r','w')
|
|
|
28 r.write("options(warn=-1)\n")
|
|
|
29 r.write("source('/Users/xuebing/galaxy-dist/tools/mytools/cdf.r')\n")
|
|
|
30 r.write("x = read.table('"+outbed+"',sep='\t')\n")
|
|
|
31 r.write("y = read.table('shuffled.dist',sep='\t')\n")
|
|
|
32 r.write("pdf('"+outplot+"')\n")
|
|
|
33 r.write("mycdf(list(log10(1+x[,ncol(x)]),log10(1+y[,ncol(y)])),'spatial distance',c('real','shuffled'),'topleft','log10 distance','')\n")
|
|
|
34 r.write("dev.off()\n")
|
|
|
35 r.close()
|
|
|
36 os.system("R --vanilla < tmp.r >"+outlog)
|
