diff cca.py @ 0:9bc0c48a027f draft default tip

Imported from capsule None
author devteam
date Mon, 19 May 2014 12:34:48 -0400
parents
children
line wrap: on
line diff
--- /dev/null	Thu Jan 01 00:00:00 1970 +0000
+++ b/cca.py	Mon May 19 12:34:48 2014 -0400
@@ -0,0 +1,158 @@
+#!/usr/bin/env python
+
+import sys, string
+from rpy import *
+import numpy
+
+def stop_err(msg):
+    sys.stderr.write(msg)
+    sys.exit()
+
+infile = sys.argv[1]
+x_cols = sys.argv[2].split(',')
+y_cols = sys.argv[3].split(',')
+
+x_scale = x_center = "FALSE"
+if sys.argv[4] == 'both':
+    x_scale = x_center = "TRUE"
+elif sys.argv[4] == 'center':
+    x_center = "TRUE"
+elif sys.argv[4] == 'scale':
+    x_scale = "TRUE"
+    
+y_scale = y_center = "FALSE"
+if sys.argv[5] == 'both':
+    y_scale = y_center = "TRUE"
+elif sys.argv[5] == 'center':
+    y_center = "TRUE"
+elif sys.argv[5] == 'scale':
+    y_scale = "TRUE"
+
+std_scores = "FALSE"   
+if sys.argv[6] == "yes":
+    std_scores = "TRUE"
+    
+outfile = sys.argv[7]
+outfile2 = sys.argv[8]
+
+fout = open(outfile,'w')
+elems = []
+for i, line in enumerate( file ( infile )):
+    line = line.rstrip('\r\n')
+    if len( line )>0 and not line.startswith( '#' ):
+        elems = line.split( '\t' )
+        break 
+    if i == 30:
+        break # Hopefully we'll never get here...
+
+if len( elems )<1:
+    stop_err( "The data in your input dataset is either missing or not formatted properly." )
+
+x_vals = []
+
+for k,col in enumerate(x_cols):
+    x_cols[k] = int(col)-1
+    x_vals.append([])
+
+y_vals = []
+
+for k,col in enumerate(y_cols):
+    y_cols[k] = int(col)-1
+    y_vals.append([])
+
+skipped = 0
+for ind,line in enumerate( file( infile )):
+    if line and not line.startswith( '#' ):
+        try:
+            fields = line.strip().split("\t")
+            valid_line = True
+            for col in x_cols+y_cols:
+                try:
+                    assert float(fields[col])
+                except:
+                    skipped += 1
+                    valid_line = False
+                    break
+            if valid_line:
+                for k,col in enumerate(x_cols):
+                    try:
+                        xval = float(fields[col])
+                    except:
+                        xval = NaN#
+                    x_vals[k].append(xval)
+                for k,col in enumerate(y_cols):
+                    try:
+                        yval = float(fields[col])
+                    except:
+                        yval = NaN#
+                    y_vals[k].append(yval)
+        except:
+            skipped += 1
+
+x_vals1 = numpy.asarray(x_vals).transpose()
+y_vals1 = numpy.asarray(y_vals).transpose()
+
+x_dat= r.list(array(x_vals1))
+y_dat= r.list(array(y_vals1))
+
+try:
+    r.suppressWarnings(r.library("yacca"))
+except:
+    stop_err("Missing R library yacca.")
+    
+set_default_mode(NO_CONVERSION)
+try:
+    xcolnames = ["c%d" %(el+1) for el in x_cols]
+    ycolnames = ["c%d" %(el+1) for el in y_cols]
+    cc = r.cca(x=x_dat, y=y_dat, xlab=xcolnames, ylab=ycolnames, xcenter=r(x_center), ycenter=r(y_center), xscale=r(x_scale), yscale=r(y_scale), standardize_scores=r(std_scores))
+    ftest = r.F_test_cca(cc)
+except RException, rex:
+    stop_err("Encountered error while performing CCA on the input data: %s" %(rex))
+
+set_default_mode(BASIC_CONVERSION)
+summary = r.summary(cc)
+
+ncomps = len(summary['corr'])
+comps = summary['corr'].keys()
+corr = summary['corr'].values()
+xlab = summary['xlab']
+ylab = summary['ylab']
+
+for i in range(ncomps):
+    corr[comps.index('CV %s' %(i+1))] = summary['corr'].values()[i]
+
+ftest=ftest.as_py()
+print >>fout, "#Component\t%s" %("\t".join(["%s" % el for el in range(1,ncomps+1)]))
+print >>fout, "#Correlation\t%s" %("\t".join(["%.4g" % el for el in corr]))
+print >>fout, "#F-statistic\t%s" %("\t".join(["%.4g" % el for el in ftest['statistic']]))
+print >>fout, "#p-value\t%s" %("\t".join(["%.4g" % el for el in ftest['p.value']]))
+
+print >>fout, "#X-Coefficients\t%s" %("\t".join(["%s" % el for el in range(1,ncomps+1)]))
+for i,val in enumerate(summary['xcoef']):
+    print >>fout, "%s\t%s" %(xlab[i], "\t".join(["%.4g" % el for el in val]))
+
+print >>fout, "#Y-Coefficients\t%s" %("\t".join(["%s" % el for el in range(1,ncomps+1)]))
+for i,val in enumerate(summary['ycoef']):
+    print >>fout, "%s\t%s" %(ylab[i], "\t".join(["%.4g" % el for el in val]))
+       
+print >>fout, "#X-Loadings\t%s" %("\t".join(["%s" % el for el in range(1,ncomps+1)]))
+for i,val in enumerate(summary['xstructcorr']):
+    print >>fout, "%s\t%s" %(xlab[i], "\t".join(["%.4g" % el for el in val]))
+
+print >>fout, "#Y-Loadings\t%s" %("\t".join(["%s" % el for el in range(1,ncomps+1)]))
+for i,val in enumerate(summary['ystructcorr']):
+    print >>fout, "%s\t%s" %(ylab[i], "\t".join(["%.4g" % el for el in val]))
+
+print >>fout, "#X-CrossLoadings\t%s" %("\t".join(["%s" % el for el in range(1,ncomps+1)]))
+for i,val in enumerate(summary['xcrosscorr']):
+    print >>fout, "%s\t%s" %(xlab[i], "\t".join(["%.4g" % el for el in val]))
+
+print >>fout, "#Y-CrossLoadings\t%s" %("\t".join(["%s" % el for el in range(1,ncomps+1)]))
+for i,val in enumerate(summary['ycrosscorr']):
+    print >>fout, "%s\t%s" %(ylab[i], "\t".join(["%.4g" % el for el in val]))
+
+r.pdf( outfile2, 8, 8 )
+#r.plot(cc)
+for i in range(ncomps):
+    r.helio_plot(cc, cv = i+1, main = r.paste("Explained Variance for CV",i+1), type = "variance")
+r.dev_off()
\ No newline at end of file