annotate scatterplot.py @ 0:ffcdde989859 draft

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