def Glm(x, y, famil = 'gaussian'):
"""
Compute the slope and intercept of the 2 given vector generalized linear regression.
:param x: X-axis values
:param y: Y-axis values
:param famil: family objects for models and link function
:type x: float list
:type y: float list
:type famil: string
:returns: the slope and the intercept of the generalized linear regression
:rtype: float
:note: the 2 vector/list must have the same size
"""
model = rpy.r("Y~X")
d = rpy.r.data_frame(X=x, Y=y)
reg = rpy.r.glm(model, data = d, family = famil)
#print reg
intercept = reg['coefficients']['(Intercept)']
slope = reg['coefficients']['X']
family = reg['family']['family']
#link = reg['link']
data = {'Intercept':intercept, 'Slope':slope, 'Family': family}#, 'Link':lin}
return data
def multiReg(x, y, colList, alpha):
#d = rpy.r.data_frame(y)
newX = []
d = {'Y':y}
model_string = "Y~"
#names = ["Y"]
for i in colList:
name = "X" + str(i)
#names.append( name )
d[name] = x[i]
newX.append(x[i])
model_string = model_string + name + "+"
print d
#rpy.r.colnames(d) = names
model_string = model_string + "1"
model = rpy.r(model_string)
n = rpy.sqrt( len( y ) )
norm = rpy.r.qnorm( 1. - ( alpha/200. ) )
Rlm = rpy.with_mode(rpy.NO_CONVERSION, rpy.r.lm)
reg = Rlm(model, data = d)
result = rpy.r.summary(reg)
print result
coef =result['coefficients']
r2 = result['r.squared']
r2adj = result['adj.r.squared']
ic = result['sigma']*norm/n
regressor = coef[:, 0]
data = {'regressor':regressor[1:], 'intercept':regressor[0], 'r2':r2, 'adj_r2':r2adj, 'ic':ic, 'x':newX, 'y':y}
return data
def regLinOri(x, y, alpha=5):
"""
Compute the slope of the 2 given vector pass-through-origin linear regression.
:Parameters:
- `x`: X-axis values
- `y`: Y-axis values
:Types:
- `x`: float list
- `y`: float list
:returns: the slope of the linear regression
:returntype: float
:attention: the 2 vector/list must have the same size
"""
model = rpy.r("Y~-1+X")
data = regression(x, y, model, alpha )
return data
def regLin(x, y, alpha=5):
"""
Compute the slope and intercept of the 2 given vector linear regression.
:Parameters:
- `x`: X-axis values
- `y`: Y-axis values
:Types:
- `x`: float list
- `y`: float list
:returns: the slope and the intercept of the linear regression
:rtype: float cople
:attention: the 2 vector/list must have the same size
"""
model = rpy.r("Y~X")
data = regression(x, y, model, alpha )
return data
def regLinOri(x, y, alpha=5):
"""
Compute the slope of the 2 given vector pass-through-origin linear regression.
:Parameters:
- `x`: X-axis values
- `y`: Y-axis values
:Types:
- `x`: float list
- `y`: float list
:returns: the slope of the linear regression
:returntype: float
:attention: the 2 vector/list must have the same size
"""
model = rpy.r("Y~-1+X")
data = regression(x, y, model, alpha)
return data
def regLin(x, y, alpha=5):
"""
Compute the slope and intercept of the 2 given vector linear regression.
:Parameters:
- `x`: X-axis values
- `y`: Y-axis values
:Types:
- `x`: float list
- `y`: float list
:returns: the slope and the intercept of the linear regression
:rtype: float cople
:attention: the 2 vector/list must have the same size
"""
model = rpy.r("Y~X")
data = regression(x, y, model, alpha)
return data
def multiReg(x, y, colList, alpha):
#d = rpy.r.data_frame(y)
newX = []
d = {'Y': y}
model_string = "Y~"
#names = ["Y"]
for i in colList:
name = "X" + str(i)
#names.append( name )
d[name] = x[i]
newX.append(x[i])
model_string = model_string + name + "+"
print d
#rpy.r.colnames(d) = names
model_string = model_string + "1"
model = rpy.r(model_string)
n = rpy.sqrt(len(y))
norm = rpy.r.qnorm(1. - (alpha / 200.))
Rlm = rpy.with_mode(rpy.NO_CONVERSION, rpy.r.lm)
reg = Rlm(model, data=d)
result = rpy.r.summary(reg)
print result
coef = result['coefficients']
r2 = result['r.squared']
r2adj = result['adj.r.squared']
ic = result['sigma'] * norm / n
regressor = coef[:, 0]
data = {
'regressor': regressor[1:],
'intercept': regressor[0],
'r2': r2,
'adj_r2': r2adj,
'ic': ic,
'x': newX,
'y': y
}
return data
def Glm(x, y, famil='gaussian'):
"""
Compute the slope and intercept of the 2 given vector generalized linear regression.
:param x: X-axis values
:param y: Y-axis values
:param famil: family objects for models and link function
:type x: float list
:type y: float list
:type famil: string
:returns: the slope and the intercept of the generalized linear regression
:rtype: float
:note: the 2 vector/list must have the same size
"""
model = rpy.r("Y~X")
d = rpy.r.data_frame(X=x, Y=y)
reg = rpy.r.glm(model, data=d, family=famil)
#print reg
intercept = reg['coefficients']['(Intercept)']
slope = reg['coefficients']['X']
family = reg['family']['family']
#link = reg['link']
data = {
'Intercept': intercept,
'Slope': slope,
'Family': family
} #, 'Link':lin}
return data
def main():
try:
datafile = sys.argv[1]
outfile_name = sys.argv[2]
expression = sys.argv[3]
except:
stop_err('Usage: python gsummary.py input_file ouput_file expression')
math_allowed = S3_METHODS()['Math']
ops_allowed = S3_METHODS()['Ops']
# Check for invalid expressions
for word in re.compile('[a-zA-Z]+').findall(expression):
if word and word not in math_allowed:
stop_err(
"Invalid expression '%s': term '%s' is not recognized or allowed"
% (expression, word))
symbols = set()
for symbol in re.compile('[^a-z0-9\s]+').findall(expression):
if symbol and symbol not in ops_allowed:
stop_err(
"Invalid expression '%s': operator '%s' is not recognized or allowed"
% (expression, symbol))
else:
symbols.add(symbol)
if len(symbols) == 1 and ',' in symbols:
# User may have entered a comma-separated list r_data_frame columns
stop_err(
"Invalid columns '%s': this tool requires a single column or expression"
% expression)
# Find all column references in the expression
cols = []
for col in re.compile('c[0-9]+').findall(expression):
try:
cols.append(int(col[1:]) - 1)
except:
pass
tmp_file = tempfile.NamedTemporaryFile('w+b')
# Write the R header row to the temporary file
hdr_str = "\t".join("c%s" % str(col + 1) for col in cols)
tmp_file.write("%s\n" % hdr_str)
skipped_lines = 0
first_invalid_line = 0
i = 0
for i, line in enumerate(open(datafile)):
line = line.rstrip('\r\n')
if line and not line.startswith('#'):
valid = True
fields = line.split('\t')
# Write the R data row to the temporary file
for col in cols:
try:
float(fields[col])
except:
skipped_lines += 1
if not first_invalid_line:
first_invalid_line = i + 1
valid = False
break
if valid:
data_str = "\t".join(fields[col] for col in cols)
tmp_file.write("%s\n" % data_str)
tmp_file.flush()
if skipped_lines == i + 1:
stop_err(
"Invalid column or column data values invalid for computation. See tool tips and syntax for data requirements."
)
else:
# summary function and return labels
set_default_mode(NO_CONVERSION)
summary_func = r(
"function( x ) { c( sum=sum( as.numeric( x ), na.rm=T ), mean=mean( as.numeric( x ), na.rm=T ), stdev=sd( as.numeric( x ), na.rm=T ), quantile( as.numeric( x ), na.rm=TRUE ) ) }"
)
headings = ['sum', 'mean', 'stdev', '0%', '25%', '50%', '75%', '100%']
headings_str = "\t".join(headings)
r_data_frame = r.read_table(tmp_file.name, header=True, sep="\t")
outfile = open(outfile_name, 'w')
for col in re.compile('c[0-9]+').findall(expression):
r.assign(col, r["$"](r_data_frame, col))
try:
summary = summary_func(r(expression))
except RException as s:
outfile.close()
stop_err("Computation resulted in the following error: %s" %
str(s))
summary = summary.as_py(BASIC_CONVERSION)
outfile.write("#%s\n" % headings_str)
if type(summary) is dict:
# using rpy
outfile.write("%s\n" %
"\t".join(["%g" % summary[k] for k in headings]))
else:
# using rpy2
outfile.write("%s\n" % "\t".join(["%g" % k for k in summary]))
outfile.close()
if skipped_lines:
print "Skipped %d invalid lines beginning with line #%d. See tool tips for data requirements." % (
skipped_lines, first_invalid_line)
def main():
try:
datafile = sys.argv[1]
outfile_name = sys.argv[2]
expression = sys.argv[3]
except Exception:
stop_err('Usage: python gsummary.py input_file ouput_file expression')
math_allowed = S3_METHODS()['Math']
ops_allowed = S3_METHODS()['Ops']
# Check for invalid expressions
for word in re.compile('[a-zA-Z]+').findall(expression):
if word and word not in math_allowed:
stop_err("Invalid expression '%s': term '%s' is not recognized or allowed" % (expression, word))
symbols = set()
for symbol in re.compile('[^a-z0-9\s]+').findall(expression):
if symbol and symbol not in ops_allowed:
stop_err("Invalid expression '%s': operator '%s' is not recognized or allowed" % (expression, symbol))
else:
symbols.add(symbol)
if len(symbols) == 1 and ',' in symbols:
# User may have entered a comma-separated list r_data_frame columns
stop_err("Invalid columns '%s': this tool requires a single column or expression" % expression)
# Find all column references in the expression
cols = []
for col in re.compile('c[0-9]+').findall(expression):
try:
cols.append(int(col[1:]) - 1)
except Exception:
pass
tmp_file = tempfile.NamedTemporaryFile('w+')
# Write the R header row to the temporary file
hdr_str = "\t".join("c%s" % str(col + 1) for col in cols)
tmp_file.write("%s\n" % hdr_str)
skipped_lines = 0
first_invalid_line = 0
i = 0
for i, line in enumerate(open(datafile)):
line = line.rstrip('\r\n')
if line and not line.startswith('#'):
valid = True
fields = line.split('\t')
# Write the R data row to the temporary file
for col in cols:
try:
float(fields[col])
except Exception:
skipped_lines += 1
if not first_invalid_line:
first_invalid_line = i + 1
valid = False
break
if valid:
data_str = "\t".join(fields[col] for col in cols)
tmp_file.write("%s\n" % data_str)
tmp_file.flush()
if skipped_lines == i + 1:
stop_err("Invalid column or column data values invalid for computation. See tool tips and syntax for data requirements.")
else:
# summary function and return labels
set_default_mode(NO_CONVERSION)
summary_func = r("function( x ) { c( sum=sum( as.numeric( x ), na.rm=T ), mean=mean( as.numeric( x ), na.rm=T ), stdev=sd( as.numeric( x ), na.rm=T ), quantile( as.numeric( x ), na.rm=TRUE ) ) }")
headings = ['sum', 'mean', 'stdev', '0%', '25%', '50%', '75%', '100%']
headings_str = "\t".join(headings)
r_data_frame = r.read_table(tmp_file.name, header=True, sep="\t")
outfile = open(outfile_name, 'w')
for col in re.compile('c[0-9]+').findall(expression):
r.assign(col, r["$"](r_data_frame, col))
try:
summary = summary_func(r(expression))
except RException as s:
outfile.close()
stop_err("Computation resulted in the following error: %s" % str(s))
summary = summary.as_py(BASIC_CONVERSION)
outfile.write("#%s\n" % headings_str)
if type(summary) is dict:
# using rpy
outfile.write("%s\n" % "\t".join(["%g" % summary[k] for k in headings]))
else:
# using rpy2
outfile.write("%s\n" % "\t".join(["%g" % k for k in summary]))
outfile.close()
if skipped_lines:
print("Skipped %d invalid lines beginning with line #%d. See tool tips for data requirements." % (skipped_lines, first_invalid_line))
