80
|
1 #!/usr/bin/env python
|
|
2
|
|
3 import sys, string
|
|
4 #from rpy import *
|
|
5 import rpy2.robjects as robjects
|
|
6 import rpy2.rlike.container as rlc
|
|
7 from rpy2.robjects.packages import importr
|
|
8 r = robjects.r
|
|
9 grdevices = importr('grDevices')
|
|
10 import numpy
|
|
11
|
|
12 def stop_err(msg):
|
|
13 sys.stderr.write(msg)
|
|
14 sys.exit()
|
|
15
|
|
16 infile = sys.argv[1]
|
|
17 x_cols = sys.argv[2].split(',')
|
|
18 method = sys.argv[3]
|
|
19 outfile = sys.argv[4]
|
|
20 outfile2 = sys.argv[5]
|
|
21
|
|
22 if method == 'svd':
|
|
23 scale = center = "FALSE"
|
|
24 if sys.argv[6] == 'both':
|
|
25 scale = center = "TRUE"
|
|
26 elif sys.argv[6] == 'center':
|
|
27 center = "TRUE"
|
|
28 elif sys.argv[6] == 'scale':
|
|
29 scale = "TRUE"
|
|
30
|
|
31 fout = open(outfile,'w')
|
|
32 elems = []
|
|
33 for i, line in enumerate( file ( infile )):
|
|
34 line = line.rstrip('\r\n')
|
|
35 if len( line )>0 and not line.startswith( '#' ):
|
|
36 elems = line.split( '\t' )
|
|
37 break
|
|
38 if i == 30:
|
|
39 break # Hopefully we'll never get here...
|
|
40
|
|
41 if len( elems )<1:
|
|
42 stop_err( "The data in your input dataset is either missing or not formatted properly." )
|
|
43
|
|
44 x_vals = []
|
|
45
|
|
46 for k,col in enumerate(x_cols):
|
|
47 x_cols[k] = int(col)-1
|
|
48 # x_vals.append([])
|
|
49
|
|
50 NA = 'NA'
|
|
51 skipped = 0
|
|
52 for ind,line in enumerate( file( infile )):
|
|
53 if line and not line.startswith( '#' ):
|
|
54 try:
|
|
55 fields = line.strip().split("\t")
|
|
56 valid_line = True
|
|
57 for k,col in enumerate(x_cols):
|
|
58 try:
|
|
59 xval = float(fields[col])
|
|
60 except:
|
|
61 skipped += 1
|
|
62 valid_line = False
|
|
63 break
|
|
64 if valid_line:
|
|
65 for k,col in enumerate(x_cols):
|
|
66 xval = float(fields[col])
|
|
67 #x_vals[k].append(xval)
|
|
68 x_vals.append(xval)
|
|
69 except:
|
|
70 skipped += 1
|
|
71
|
|
72 #x_vals1 = numpy.asarray(x_vals).transpose()
|
|
73 #dat= r.list(array(x_vals1))
|
|
74 dat = r['matrix'](robjects.FloatVector(x_vals),ncol=len(x_cols),byrow=True)
|
|
75
|
|
76 #set_default_mode(NO_CONVERSION)
|
|
77 try:
|
|
78 if method == "cor":
|
|
79 #pc = r.princomp(r.na_exclude(dat), cor = r("TRUE"))
|
|
80 pc = r.princomp(r['na.exclude'](dat), cor = r("TRUE"))
|
|
81 elif method == "cov":
|
|
82 #pc = r.princomp(r.na_exclude(dat), cor = r("FALSE"))
|
|
83 pc = r.princomp(r['na.exclude'](dat), cor = r("FALSE"))
|
|
84 elif method=="svd":
|
|
85 #pc = r.prcomp(r.na_exclude(dat), center = r(center), scale = r(scale))
|
|
86 pc = r.prcomp(r['na.exclude'](dat), center = r(center), scale = r(scale))
|
|
87 #except RException, rex:
|
|
88 except Exception, rex: # need to find rpy2 RException
|
|
89 stop_err("Encountered error while performing PCA on the input data: %s" %(rex))
|
|
90
|
|
91 #set_default_mode(BASIC_CONVERSION)
|
|
92 summary = r.summary(pc, loadings="TRUE")
|
|
93 #ncomps = len(summary['sdev'])
|
|
94 ncomps = len(summary.rx2('sdev'))
|
|
95
|
|
96 #if type(summary['sdev']) == type({}):
|
|
97 # comps_unsorted = summary['sdev'].keys()
|
|
98 # comps=[]
|
|
99 # sd = summary['sdev'].values()
|
|
100 # for i in range(ncomps):
|
|
101 # sd[i] = summary['sdev'].values()[comps_unsorted.index('Comp.%s' %(i+1))]
|
|
102 # comps.append('Comp.%s' %(i+1))
|
|
103 #elif type(summary['sdev']) == type([]):
|
|
104 # comps=[]
|
|
105 # for i in range(ncomps):
|
|
106 # comps.append('Comp.%s' %(i+1))
|
|
107 # sd = summary['sdev']
|
|
108
|
|
109 comps=[]
|
|
110 for i in range(ncomps):
|
|
111 comps.append('Comp.%s' %(i+1))
|
|
112 sd = summary.rx2('sdev')
|
|
113
|
|
114 print >>fout, "#Component\t%s" %("\t".join(["%s" % el for el in range(1,ncomps+1)]))
|
|
115 #print >>fout, "#Std. deviation\t%s" %("\t".join(["%.4g" % el for el in sd]))
|
|
116 print >>fout, "#Std. deviation\t%s" %("\t".join(["%.4g" % el for el in sd]))
|
|
117 total_var = 0
|
|
118 vars = []
|
|
119 for s in sd:
|
|
120 var = s*s
|
|
121 total_var += var
|
|
122 vars.append(var)
|
|
123 for i,var in enumerate(vars):
|
|
124 vars[i] = vars[i]/total_var
|
|
125
|
|
126 print >>fout, "#Proportion of variance explained\t%s" %("\t".join(["%.4g" % el for el in vars]))
|
|
127
|
|
128 print >>fout, "#Loadings\t%s" %("\t".join(["%s" % el for el in range(1,ncomps+1)]))
|
|
129 xcolnames = ["c%d" %(el+1) for el in x_cols]
|
|
130 #if 'loadings' in summary: #in case of princomp
|
|
131 if 'loadings' in summary.names: #in case of princomp
|
|
132 loadings = 'loadings'
|
|
133 #elif 'rotation' in summary: #in case of prcomp
|
|
134 elif 'rotation' in summary.names: #in case of prcomp
|
|
135 loadings = 'rotation'
|
|
136 #for i,val in enumerate(summary[loadings]):
|
|
137 # print >>fout, "%s\t%s" %(xcolnames[i], "\t".join(["%.4g" % el for el in val]))
|
|
138 vm = summary.rx2(loadings)
|
|
139 for i in range(vm.nrow):
|
|
140 vals = []
|
|
141 for j in range(vm.ncol):
|
|
142 vals.append("%.4g" % vm.rx2(i+1,j+1)[0])
|
|
143 print >>fout, "%s\t%s" %(xcolnames[i], "\t".join(vals))
|
|
144
|
|
145 print >>fout, "#Scores\t%s" %("\t".join(["%s" % el for el in range(1,ncomps+1)]))
|
|
146 #if 'scores' in summary: #in case of princomp
|
|
147 if 'scores' in summary.names: #in case of princomp
|
|
148 scores = 'scores'
|
|
149 #elif 'x' in summary: #in case of prcomp
|
|
150 elif 'x' in summary.names: #in case of prcomp
|
|
151 scores = 'x'
|
|
152 #for obs,sc in enumerate(summary[scores]):
|
|
153 # print >>fout, "%s\t%s" %(obs+1, "\t".join(["%.4g" % el for el in sc]))
|
|
154 vm = summary.rx2(scores)
|
|
155 for i in range(vm.nrow):
|
|
156 vals = []
|
|
157 for j in range(vm.ncol):
|
|
158 vals.append("%.4g" % vm.rx2(i+1,j+1)[0])
|
|
159 print >>fout, "%s\t%s" %(i+1, "\t".join(vals))
|
|
160 r.pdf( outfile2, 8, 8 )
|
|
161 r.biplot(pc)
|
|
162 #r.dev_off()
|
|
163 grdevices.dev_off()
|
|
164
|