|
0
|
1 #!/usr/bin/env python
|
|
|
2
|
|
|
3 import sys, string
|
|
|
4 import rpy2.robjects as robjects
|
|
|
5 import rpy2.rlike.container as rlc
|
|
|
6 from rpy2.robjects.packages import importr
|
|
|
7 r = robjects.r
|
|
|
8 grdevices = importr('grDevices')
|
|
|
9 # from rpy import *
|
|
|
10 import numpy
|
|
|
11
|
|
|
12
|
|
|
13 def stop_err(msg):
|
|
|
14 sys.stderr.write(msg)
|
|
|
15 sys.exit()
|
|
|
16
|
|
|
17 infile = sys.argv[1]
|
|
|
18 y_col = int(sys.argv[2])-1
|
|
|
19 x_cols = sys.argv[3].split(',')
|
|
|
20 outfile = sys.argv[4]
|
|
|
21 outfile2 = sys.argv[5]
|
|
|
22
|
|
|
23 print "Predictor columns: %s; Response column: %d" %(x_cols,y_col+1)
|
|
|
24 fout = open(outfile,'w')
|
|
|
25 elems = []
|
|
|
26 for i, line in enumerate( file ( infile )):
|
|
|
27 line = line.rstrip('\r\n')
|
|
|
28 if len( line )>0 and not line.startswith( '#' ):
|
|
|
29 elems = line.split( '\t' )
|
|
|
30 break
|
|
|
31 if i == 30:
|
|
|
32 break # Hopefully we'll never get here...
|
|
|
33
|
|
|
34 if len( elems )<1:
|
|
|
35 stop_err( "The data in your input dataset is either missing or not formatted properly." )
|
|
|
36
|
|
|
37 y_vals = []
|
|
|
38 x_vals = []
|
|
|
39
|
|
|
40 for k,col in enumerate(x_cols):
|
|
|
41 x_cols[k] = int(col)-1
|
|
|
42 # x_vals.append([])
|
|
|
43
|
|
|
44 NA = 'NA'
|
|
|
45 for ind,line in enumerate( file( infile )):
|
|
|
46 if line and not line.startswith( '#' ):
|
|
|
47 try:
|
|
|
48 fields = line.split("\t")
|
|
|
49 try:
|
|
|
50 yval = float(fields[y_col])
|
|
|
51 except:
|
|
|
52 yval = r('NA')
|
|
|
53 y_vals.append(yval)
|
|
|
54 for k,col in enumerate(x_cols):
|
|
|
55 try:
|
|
|
56 xval = float(fields[col])
|
|
|
57 except:
|
|
|
58 xval = r('NA')
|
|
|
59 # x_vals[k].append(xval)
|
|
|
60 x_vals.append(xval)
|
|
|
61 except:
|
|
|
62 pass
|
|
|
63 # x_vals1 = numpy.asarray(x_vals).transpose()
|
|
|
64 # dat= r.list(x=array(x_vals1), y=y_vals)
|
|
|
65 fv = robjects.FloatVector(x_vals)
|
|
|
66 m = r['matrix'](fv, ncol=len(x_cols),byrow=True)
|
|
|
67 # ensure order for generating formula
|
|
|
68 od = rlc.OrdDict([('y',robjects.FloatVector(y_vals)),('x',m)])
|
|
|
69 dat = robjects.DataFrame(od)
|
|
|
70 # convert dat.names: ["y","x.1","x.2"] to formula string: 'y ~ x.1 + x.2'
|
|
|
71 formula = ' + '.join(dat.names).replace('+','~',1)
|
|
|
72
|
|
|
73 #set_default_mode(NO_CONVERSION)
|
|
|
74 try:
|
|
|
75 #linear_model = r.lm(r("y ~ x"), data = r.na_exclude(dat))
|
|
|
76 linear_model = r.lm(formula, data = r['na.exclude'](dat))
|
|
|
77 except RException, rex:
|
|
|
78 stop_err("Error performing linear regression on the input data.\nEither the response column or one of the predictor columns contain only non-numeric or invalid values.")
|
|
|
79 #set_default_mode(BASIC_CONVERSION)
|
|
|
80
|
|
|
81 #coeffs=linear_model.as_py()['coefficients']
|
|
|
82 #yintercept= coeffs['(Intercept)']
|
|
|
83 coeffs=linear_model.rx2('coefficients')
|
|
|
84 yintercept= coeffs.rx2('(Intercept)')[0]
|
|
|
85 summary = r.summary(linear_model)
|
|
|
86
|
|
|
87 #co = summary.get('coefficients', 'NA')
|
|
|
88 co = summary.rx2("coefficients")
|
|
|
89
|
|
|
90 """
|
|
|
91 if len(co) != len(x_vals)+1:
|
|
|
92 stop_err("Stopped performing linear regression on the input data, since one of the predictor columns contains only non-numeric or invalid values.")
|
|
|
93 """
|
|
|
94 #print >>fout, "p-value (Y-intercept)\t%s" %(co[0][3])
|
|
|
95 print >>fout, "p-value (Y-intercept)\t%s" %(co.rx(1,4)[0])
|
|
|
96
|
|
|
97 if len(x_vals) == 1: #Simple linear regression case with 1 predictor variable
|
|
|
98 try:
|
|
|
99 #slope = coeffs['x']
|
|
|
100 slope = r.round(float(coeffs.rx2('x')[0]), digits=10)
|
|
|
101 except:
|
|
|
102 slope = 'NA'
|
|
|
103 try:
|
|
|
104 #pval = co[1][3]
|
|
|
105 pval = r.round(float(co.rx(2,4)[0]), digits=10)
|
|
|
106 except:
|
|
|
107 pval = 'NA'
|
|
|
108 print >>fout, "Slope (c%d)\t%s" %(x_cols[0]+1,slope)
|
|
|
109 print >>fout, "p-value (c%d)\t%s" %(x_cols[0]+1,pval)
|
|
|
110 else: #Multiple regression case with >1 predictors
|
|
|
111 ind=1
|
|
|
112 #while ind < len(coeffs.keys()):
|
|
|
113 while ind < len(coeffs.names):
|
|
|
114 # print >>fout, "Slope (c%d)\t%s" %(x_cols[ind-1]+1,coeffs['x'+str(ind)])
|
|
|
115 print >>fout, "Slope (c%d)\t%s" %(x_cols[ind-1]+1,coeffs.rx2(coeffs.names[ind])[0])
|
|
|
116 try:
|
|
|
117 #pval = co[ind][3]
|
|
|
118 pval = r.round(float(co.rx(ind+1,4)[0]), digits=10)
|
|
|
119 except:
|
|
|
120 pval = 'NA'
|
|
|
121 print >>fout, "p-value (c%d)\t%s" %(x_cols[ind-1]+1,pval)
|
|
|
122 ind+=1
|
|
|
123
|
|
|
124 rsq = summary.rx2('r.squared')[0]
|
|
|
125 adjrsq = summary.rx2('adj.r.squared')[0]
|
|
|
126 fstat = summary.rx2('fstatistic').rx2('value')[0]
|
|
|
127 sigma = summary.rx2('sigma')[0]
|
|
|
128
|
|
|
129 try:
|
|
|
130 rsq = r.round(float(rsq), digits=5)
|
|
|
131 adjrsq = r.round(float(adjrsq), digits=5)
|
|
|
132 fval = r.round(fstat['value'], digits=5)
|
|
|
133 fstat['value'] = str(fval)
|
|
|
134 sigma = r.round(float(sigma), digits=10)
|
|
|
135 except:
|
|
|
136 pass
|
|
|
137
|
|
|
138 print >>fout, "R-squared\t%s" %(rsq)
|
|
|
139 print >>fout, "Adjusted R-squared\t%s" %(adjrsq)
|
|
|
140 print >>fout, "F-statistic\t%s" %(fstat)
|
|
|
141 print >>fout, "Sigma\t%s" %(sigma)
|
|
|
142
|
|
|
143 r.pdf( outfile2, 8, 8 )
|
|
|
144 if len(x_vals) == 1: #Simple linear regression case with 1 predictor variable
|
|
|
145 sub_title = "Slope = %s; Y-int = %s" %(slope,yintercept)
|
|
|
146 try:
|
|
|
147 r.plot(x=x_vals[0], y=y_vals, xlab="X", ylab="Y", sub=sub_title, main="Scatterplot with regression")
|
|
|
148 r.abline(a=yintercept, b=slope, col="red")
|
|
|
149 except:
|
|
|
150 pass
|
|
|
151 else:
|
|
|
152 r.pairs(dat, main="Scatterplot Matrix", col="blue")
|
|
|
153 try:
|
|
|
154 r.plot(linear_model)
|
|
|
155 except:
|
|
|
156 pass
|
|
|
157 #r.dev_off()
|
|
|
158 grdevices.dev_off()
|
