def fusegroupcolumns(tabdata, group):
"""
function which merges the columns
in the group statement into
one column with fused name
this allows to group later
in ggplot2
"""
if isinstance(tabdata, list):
tmpdata = fhutils.Table()
tmpdata.columnames = tabdata[0]
tmpdata.data = tabdata[1:]
tabdata = tmpdata.iclone()
newcolumnames = list(tabdata.columnames)
groupjoin = ''.join([str(x) for x in group])
newcolumnames.append(groupjoin)
ntab = fhutils.Table()
ntab.columnames = newcolumnames
for row in tabdata.iterrows():
mergecolumn = ''.join([str(row[ele]) for ele in group])
newrowlist = [row[ele] for ele in tabdata.columnames] + [mergecolumn]
ntab.data.append(newrowlist)
return ntab, groupjoin
def pyobj2dataframe(valuelis, columnames=True, rownames=False):
"""
create rpy2-R data frame from table or list
to be used in R
"""
if isinstance(valuelis, robjects.vectors.DataFrame):
return valuelis
elif isinstance(valuelis, fhutils.Table):
pytab = fhutils.Table('data')
assert '-' not in valuelis.columnames, valuelis.columnames
pytab.columnames = valuelis.columnames
pytab.data = valuelis.data
else:
pytab = fhutils.Table('data')
if columnames: # if columnames provided extracted from the list
assert '-' not in valuelis[0], valuelis[0]
pytab.columnames = valuelis[0]
pytab.data = valuelis[1:]
else: # add arbitrary columnames
pytab.columnames = ['v' + str(i) for i in range(0, len(valuelis[0]))]
pytab.data = valuelis
tablist = []
for col in pytab.columnames:
coldata = pytab.getcolumn(col)
nv = []
for ele in coldata:
if isinstance(ele, list) or isinstance(ele, tuple):
ele = ele[0]
if isinstance(ele, int) or isinstance(ele, float):
nv.append(float(ele))
elif ele.isdigit():
nv.append(float(ele))
elif ele == 'NA':
nv.append(rinterface.NARealType())
else:
nv.append(fhutils.is_numeric(ele)) # try to convert into most likely type
if sum([1 for ele in nv if isinstance(ele, float) or isinstance(ele, int)]) < len(nv):
tablist.append((col, robjects.vectors.StrVector(nv)))
else:
tablist.append((col, robjects.vectors.FloatVector(nv)))
dataframe = robjects.DataFrame(rlc.OrdDict(tablist))
if rownames:
dfrmgen = robjects.r('data.frame')
dataframe = dfrmgen(dataframe, row_names=1)
return dataframe
def dataframe2pyobj(dataframe):
table = fhutils.Table()
colnames = list(dataframe.colnames)
rownames = list(dataframe.rownames)
col2data = []
for cn, col in dataframe.items():
if isinstance(col, robjects.vectors.FactorVector) is True:
colevel = tuple(col.levels)
col = tuple(col)
ncol = []
for i in col:
k = i - 1
ncol.append(colevel[k])
else:
ncol = tuple(col)
col2data.append((cn, ncol))
col2data.append(('rownames', rownames))
col2data = dict(col2data)
table.columnames = ['rownames'] + colnames
for cname in table.columnames:
table.data.append(tuple(col2data[cname]))
table.data = fhutils.pytranspose(table.data)
return table
def pyrnparcomp(table,measurevar,groupvar):
rconsole=rpystatinit()
rnparcomp = rconsole("rnparcomp")
dataframe = pyobj2dataframe(table)
results = rmultcomp(data=dataframe, measurevar=measurevar,groupvar=groupvar)
results = dataframe2pyobj(results)
newtab = fhutils.Table('nparcomp')
newtab.columnames = ['group1', 'group2', 'p-value']
for row in results:
print(row)
return newtab
def pyrmultcomp(table,measurevar,groupvar,contrastsvector = 'none'):
rconsole=rpystatinit()
rmultcomp = rconsole("rmultcomp")
dataframe = pyobj2dataframe(table)
contrastsvector = tuple([ "{tu1} - {tu2} = 0".format(tu1 = tu[0],tu2=tu[1]) for tu1,tu2 in contrastsvector])
contrastsvector = robjects.vectors.StrVector(contrastsvector)
results = rmultcomp(data=dataframe, measurevar=measurevar,groupvar=groupvar, contrastsvector=contrastsvector)
results = dataframe2pyobj(results)
newtab = fhutils.Table('multcomp')
newtab.columnames = ['group1', 'group2', 'p-value']
for row in results:
print(row)
return newtab
def pykruskalwallis(table, formula, term, filename='kruskalimage', gformat='pdf'):
rconsole=rpystatinit()
rkruskal = rconsole("rkruskal")
dataframe = pyobj2dataframe(table)
results = rkruskal(data=dataframe, formulastring=formula, term=term)
results = dataframe2pyobj(results)
newtab = fhutils.Table('kruskal')
newtab.columnames = ['group1', 'group2', 'p-value']
for row in results:
for i in range(1, len(results.columnames)):
newtab.append((row[0], results.columnames[i], row[i]))
return newtab
def propagatetable(datatable, groupcolumn, measurecolumn, errorcolumn):
"""
primitive aggregate i.e. averaging function
for a table with error values
propagates error using ufloat
mechanism
NA filtered out by default - no control
"""
if isinstance(groupcolumn, str):
groupcolumn = (groupcolumn,)
if isinstance(groupcolumn, unicode):
groupcolumn = (groupcolumn,)
keyvalue = []
for row in datatable.iterrows():
tmp = []
for col in groupcolumn:
tmp.append(row[col])
if row[measurecolumn] in ('NA',):
continue
elif row[errorcolumn] in ('NA',):
value = row[measurecolumn]
else:
value = ufloat(row[measurecolumn], row[errorcolumn])
keyvalue.append((tuple(tmp), value))
keyvalue = fhutils.keylis2dic(keyvalue)
resultable = fhutils.Table('propagate')
resultable.columnames = list(groupcolumn) + [measurecolumn, errorcolumn]
for k, valuelist in keyvalue.items():
valuelist = [ele for ele in valuelist if ele != 'NA'] # NA filtered out
ufavg = ufmeanstd(valuelist, asufloat=True)
row = list(k)
if isinstance(ufavg, uncertainties.UFloat):
row += [ufavg.nominal_value, ufavg.std_dev]
else:
row += [ufavg, 0]
resultable.append(row)
resultable.sortrow()
return resultable