comparison scatterplot.py @ 0:ffcdde989859 draft

Uploaded
author iuc
date Tue, 29 Jul 2014 06:30:45 -0400
parents
children
comparison
equal deleted inserted replaced
-1:000000000000 0:ffcdde989859
1 #!/usr/bin/env python
2 #Greg Von Kuster
3
4 import sys
5 #from rpy import *
6 import rpy2.robjects as robjects
7 from rpy2.robjects.packages import importr
8 r = robjects.r
9 grdevices = importr('grDevices')
10
11 def stop_err(msg):
12 sys.stderr.write(msg)
13 sys.exit()
14
15 def main():
16
17 in_fname = sys.argv[1]
18 out_fname = sys.argv[2]
19 try:
20 columns = int( sys.argv[3] ) - 1, int( sys.argv[4] ) - 1
21 except:
22 stop_err( "Columns not specified, your query does not contain a column of numerical data." )
23 title = sys.argv[5]
24 xlab = sys.argv[6]
25 ylab = sys.argv[7]
26
27 matrix = []
28 skipped_lines = 0
29 first_invalid_line = 0
30 invalid_value = ''
31 invalid_column = 0
32 i = 0
33 for i, line in enumerate( file( in_fname ) ):
34 valid = True
35 line = line.rstrip( '\r\n' )
36 if line and not line.startswith( '#' ):
37 row = []
38 fields = line.split( "\t" )
39 for column in columns:
40 try:
41 val = fields[column]
42 if val.lower() == "na":
43 row.append( float( "nan" ) )
44 else:
45 row.append( float( fields[column] ) )
46 except:
47 valid = False
48 skipped_lines += 1
49 if not first_invalid_line:
50 first_invalid_line = i + 1
51 try:
52 invalid_value = fields[column]
53 except:
54 invalid_value = ''
55 invalid_column = column + 1
56 break
57 else:
58 valid = False
59 skipped_lines += 1
60 if not first_invalid_line:
61 first_invalid_line = i+1
62
63 if valid:
64 #matrix.append( row )
65 matrix += row
66
67 if skipped_lines < i:
68 try:
69 fv = robjects.FloatVector(matrix)
70 m = r['matrix'](fv, ncol=len(columns),byrow=True)
71 r.pdf( out_fname, 8, 8 )
72 #r.plot( array( matrix ), type="p", main=title, xlab=xlab, ylab=ylab, col="blue", pch=19 )
73 r.plot( m, type="p", main=title, xlab=xlab, ylab=ylab, col="blue", pch=19 )
74 #r.dev_off()
75 grdevices.dev_off()
76 except Exception, exc:
77 stop_err( "%s" %str( exc ) )
78 else:
79 stop_err( "All values in both columns %s and %s are non-numeric or empty." % ( sys.argv[3], sys.argv[4] ) )
80
81 print "Scatter plot on columns %s, %s. " % ( sys.argv[3], sys.argv[4] )
82 if skipped_lines > 0:
83 print "Skipped %d lines starting with line #%d, value '%s' in column %d is not numeric." % ( skipped_lines, first_invalid_line, invalid_value, invalid_column )
84
85 #r.quit( save="no" )
86
87 if __name__ == "__main__":
88 main()