def make_dataframe(data, xid, yid, factorid):
ns = []
steps = []
runtimes = []
for (n, v1) in data.items():
for (step, runtime) in v1.items():
ns.append(n)
steps.append(step)
runtimes.append(runtime)
df = ro.DataFrame({xid : ro.IntVector(ns),
factorid : ro.StrVector(steps),
yid : ro.IntVector(runtimes)})
return df
def calculate(self, msg):
save_path, all_predicted_time = msg
x, y, _ = plt.hist(list(itertools.chain.from_iterable(all_predicted_time)), bins=5)
qt = importr("qualityTools")
grdevices = importr('grDevices')
ytp = [(y[i]/100, y[i+1]/100) for i in range(len(y)-1)]
x_rp = []
for i in range(len(x)):
x_rp.extend([str(ytp[i])]*int(x[i]))
grdevices.png(file="%s.png" % save_path, width=600, height=400)
qt.paretoChart(robjects.StrVector(x_rp), main="predicted_time--"+save_path.basename())
grdevices.dev_off()
def heatmap(Arguments):
'''
'''
Data = load_data(Arguments)
if Arguments.OrderBy: Data = reorder_data(Data, Arguments)
#This is specifically the part you want plotted on the heatmap, NOT the sidebars
Features, Variates = get_heatmap_data(Data, Arguments)
Features = [Feature[:Feature.index(":")] for Feature in Features]
ClusteredMatrix = ro.r["hierarchical_cluster_distance_matrix"](
float_matrix(Data.Labels, Features, Variates, Arguments))
Columns = list(ClusteredMatrix.colnames)
Rows = list(ClusteredMatrix.rownames)
Matrix = np.array(ClusteredMatrix)
ZMatrix = z_matrix(Matrix)
ColorPalette = list(ro.r["colorRampPalette"](ro.StrVector(
["blue", "white", "red"]))(100))
ColorMatrix = [float2color(Vector, ColorPalette) for Vector in ZMatrix]
exit()
Heatmap = float_matrix(Data.Labels, Features, Variates, Arguments)
Legend = False #overwrite if ColSideColors is called
if "colsidecolors" in [Option.lower() for Option in Arguments.Heatmap]:
ColorFeatures, ColSideMatrix, Legend = get_colsidematrix(
Data, Arguments)
ColSideMatrix = transpose_matrix(
string_matrix(Data.Labels, ColorFeatures, ColSideMatrix,
Arguments))
else:
ColSideMatrix = False
if "rowsidecolors" in [Option.lower() for Option in Arguments.Heatmap]:
RowSideMatrix = get_rowsidematrix(Features, Arguments)
else:
RowSideMatrix = False
ro.r['pdf'](Arguments.Filename + '.pdf')
if Legend: ro.r['par'](mar=ro.IntVector([1, 1, 1, 1]))
ro.r['par'](**{'cex.axis': 0.8})
get_heatmap(Heatmap,
ColSideMatrix=ColSideMatrix,
RowSideMatrix=RowSideMatrix,
Legend=Legend)
ro.r['dev.off']()
return
def get_mail_corpus(nlon_cleaning=False):
if (nlon_cleaning):
nlon, nlon_model = training_nlon()
#Path to mail's corpus
corpus_file = 'data/mailcorpus.json'
with open(corpus_file) as data_file:
corpus = json.load(data_file)
print('Reading and cleaning emails corpus. Number of emails: ' + str(len(corpus)))
dict = {}
n = 0
#Text cleaning
for d in corpus:
if d['type_of_recipient'] == 'From':
# if not d['is_response_of'] == None:
res = EmailReplyParser.read(d['message_body'].replace('\\n', '\n'))
text = res.reply
# else:
# text = d['message_body'].replace('\\n', '\n')
n += 1
if (nlon_cleaning):
try:
soup = BS4(text, 'html.parser')
clean_message_body = soup.text
except Exception as e:
print('Error with BS4 on text:\n\n%s\n\n' % text, str(e))
clean_message_body = text.strip()
message_by_lines = text.splitlines()
list_length = len(message_by_lines)
index = 0
for count in range(0, list_length):
text1 = robjects.StrVector([message_by_lines[index]])
if nlon.NLoNPredict(nlon_model, text1)[0] == 'Not':
del message_by_lines[index]
else:
index = index + 1
clean_message_body = '\n'.join(message_by_lines)
text = clean_message_body
if not text == '':
if d['email_address'] in dict:
dict[d['email_address']].append(text)
else:
dict[d['email_address']] = [text]
print(str(n)+'/'+str(len(corpus))+'\n' if n%50==0 else '', end='')
print('Mails retrieved: '+ str(n))
print('Email addresses: '+ str(len(dict)))
return dict
def make_rvector(col, ct=COLTYPE.FLOAT):
"""Make and return an R vector for data in `col` of COLTYPE ct.
Returns:
robjects.Vector
Raises:
TypeError if the type is unknown
TypeError if it is COLTYPE.DATE but not parseable
"""
if ct == COLTYPE.INT:
vec = robjects.IntVector(col)
elif ct == COLTYPE.FLOAT:
vec = robjects.FloatVector(col)
elif ct == COLTYPE.STR:
# Use I() from R.base library to avoid conversion
# into a factor. Usually though a factor is what you want.
vec = base.I(robjects.StrVector(col))
elif ct == COLTYPE.BOOL:
vec = robjects.BoolVector(col)
elif ct == COLTYPE.FACTOR:
# conversion will happen automatically
vec = robjects.StrVector(col)
elif ct == COLTYPE.DATE:
field = col[0]
if isinstance(field, datetime.datetime):
tcol = map(datetime_to_sec, col)
elif isinstance(field, float):
tcol = col
else:
raise TypeError("Bad date type '%s' for column %d, '%s'. "
"Expected time.struct_time, "
"datetime.datetime, or float." % (
type(field), i, colnames[i]))
vec = robjects.FloatVector(tcol)
else:
raise TypeError("Unknown type '%s' for column %d, '%s'." % (
type(field), i, colnames[i]))
return(vec)
def get_dps(self, entries):
""" Wrapper to retrieve dotproducts from DPS database (neuronlistfh)
as neuronslist.
Parameters
----------
entries : int | str | list thereof, optional
Neurons to extract from DPS database. Can be:
1. int: e.g. ``hits=5`` for top 5 hits
2 .list of ints: e.g. ``hits=[2,5]`` to plot hits 2 and 5
3. string: e.g. ``hits = 'THMARCM-198F_seg1'`` to plot this neuron
4. list of strings:
e.g. ``['THMARCM-198F_seg1', npfMARCM-'F000003_seg002']``
to plot multiple neurons by their gene name
Returns
-------
neuronlist of dotproduct neurons
"""
if isinstance(entries, int):
return self.db.rx(
robjects.StrVector(self.results.ix[:entries -
1].gene_name.tolist()))
elif isinstance(entries, str):
return self.db.rx(entries)
elif isinstance(entries, (list, np.ndarray)):
if isinstance(entries[0], int):
return self.db.rx(
robjects.StrVector(
self.results.ix[entries].gene_name.tolist()))
elif isinstance(entries[0], str):
return self.db.rx(robjects.StrVector(entries))
else:
logger.error('Unable to intepret entries provided. See '
'help(NBLASTresults.plot3d) for details.')
return None
def creatruleset(self, rules):
flag = False
for rule in rules.keys():
sets = str.split(rule, '=>')
robjects.globalenv["l"] = robjects.StrVector(
re.findall("[a-z]+", sets[0]))
robjects.globalenv["r"] = robjects.StrVector(
re.findall("[a-z]+", sets[1]))
quality = rules[rule]
# robjects.globalenv["q"]=robjects.DataFrame({'support':int(quality['sup']),'confidence':float(quality['conf']),'w_Kulc':float(quality['h_Kulc'])})
robjects.globalenv["q"] = robjects.DataFrame({
'support':
int(quality['sup']),
'confidence':
float(quality['conf']),
'lift':
float(quality['lift'])
})
robjects.r('''
lm<-matrix(1,ncol=length(l))
dimnames(lm)<-list(NULL,l)
x<-matrix(0,ncol=length(r))
dimnames(x)<-list(NULL,r)
lhs<-as(cbind(lm,x),"itemMatrix")
rm<-matrix(1,ncol=length(r))
dimnames(rm)<-list(NULL,r)
x<-matrix(0,ncol=length(l))
dimnames(x)<-list(NULL,l)
rhs<-as(cbind(x,rm),"itemMatrix")
''')
if flag:
robjects.r('''
rule<-new("rules",lhs=lhs,rhs=rhs,quality=q)
ruleset<-c(ruleset,rule)
''')
else:
robjects.r('ruleset<-new("rules",lhs=lhs,rhs=rhs,quality=q)')
flag = True
def mplotHis(moptions):
perclist, rankgrouplist, rankperclist, split_points, myRankStr = group_rank(
moptions)
figname = moptions["FileID"]
mresfolder = moptions['outFolder']
ggplot = importr('ggplot2')
importr('gridExtra')
spvector = robjects.IntVector(split_points)
rankstrvector = robjects.StrVector(myRankStr)
moptions['CaseSizes'].sort()
csvector = robjects.IntVector(moptions['CaseSizes'])
#mdfperc = robjects.DataFrame({"MixedPerc":robjects.FactorVector(robjects.FloatVector(perclist), levels=percvector, labels=percvector), "Rank":robjects.FactorVector(robjects.StrVector(rankgrouplist), levels=rankstrvector, labels=rankstrvector), "Fraction":robjects.FloatVector(rankperclist)})
mdfperc = robjects.DataFrame({
"MixedPerc":
robjects.FactorVector(robjects.IntVector(perclist),
levels=csvector,
labels=csvector),
"Percentile":
robjects.FactorVector(robjects.StrVector(rankgrouplist),
levels=rankstrvector,
labels=rankstrvector),
"Fraction":
robjects.FloatVector(rankperclist)
})
robjects.r(resource_string(__name__, 'Rscript/Hist_sim_plot.R'))
robjects.r('pdf("' + mresfolder + '/hist2_' + figname + '.pdf", width=' +
("%.0f" % (len(moptions["CaseSizes"]) * 0.8)) +
', height=4, onefile = TRUE)')
robjects.globalenv['Hist_sim_plot'](mdfperc, spvector, rankstrvector)
robjects.r('dev.off()')
def plot_module_eigengene(self, module):
'''
barchart illustrating module eigengene
'''
eigengene = self.eigengenes.get_module_eigengene(module)
params = {}
params['height'] = base().as_numeric(eigengene)
limit = max(abs(base().max(eigengene)[0]), abs(base().min(eigengene)[0]))
ylim = [-1 * limit, limit]
params['ylim'] = ro.IntVector(ylim)
colors = ["red" if e[0] > 0 else "blue" for e in eigengene]
params['col'] = ro.StrVector(colors)
params['border'] = ro.NA_Logical
params['las'] = 2
params['names.arg'] = ro.StrVector(self.eigengenes.samples())
params['cex.names'] = 0.6
params['main'] = "Eigengene: " + module
manager = RManager(eigengene, params)
manager.barchart()
def module_eigengenes(self, membership):
'''
wrapper for moduleEigengenes function
calculates eigengenes from profiles &
module membership (gene -> membership dict)
'''
params = {}
params['softPower'] = self.params[
'power'] if 'power' in self.params else 6
params['expr'] = base().as_data_frame(self.transpose_data())
params['colors'] = ro.StrVector(membership)
return wgcna().moduleEigengenes(**params)
def _filter_and_values_to_RList(d):
"""`d` is a dictionary of filters: values. Returns a StrVector and
a ListVector of StrVectors"""
# Could use ListVector directly with the dict, but want to guarantee
# positional order of filters and values
f = robjects.StrVector(list(d.keys()))
v = robjects.ListVector(
rpy2.rlike.container.TaggedList(
d.values(),
tags=list(d.keys())
)
)
return f, v
def read_target_group_of_interest(targetspath):
"""
Reads a tsv file, extracts the first column (while assuming they
are transcript/gene identifiers) and sets them as a vector in R
global environament.
"""
genes = []
with open(targetspath) as fh:
for line in fh:
token = line.split("\t")
genes.append(token[0].strip())
robjects.globalenv["targetset"] = robjects.StrVector(genes)
def _run(self, cdata):
# Execute script using R
r_json_arguments = robjects.StrVector([json.dumps(cdata, indent=4)])
r_arguments = importr("rjson").fromJSON(r_json_arguments)
source_r_file(self.r_file)
result = str(r['do.call'](
'run', r_arguments)).encode("utf-8").decode("unicode_escape")
# Interpret results as string
match = R_STRING_RE.match(result)
if not match:
err_msg = "Expected a string as return value from R script. Got: {}"
raise ValueError(err_msg.format(result))
return match.groupdict()["str"]
def catch_dataset_ids(self, data_frame, id_label):
id_pos = get_item_pos(r['names'](data_frame), id_label)
ids_idx = [
item.replace('"', '')
for item in list(r['as.vector'](data_frame[id_pos]))
]
keep = [
item for item in list(r['names'](data_frame)) if item != id_label
]
keep = robjects.StrVector(keep)
data_frame_x = data_frame.rx(True, keep)
return [data_frame_x, ids_idx]
def genResampleData(infile, outfile):
'''
Resample the data n-times with replacement - generates
n flat files which are then propagated at later stages.
Files are generally small though.
'''
time_agg = list(TIME.__dict__['track2groups'].keys())
time_points = [int(str(x).split("-")[1]) for x in time_agg]
time_points.sort()
time_points = list(set(time_points))
rep_agg = list(REPLICATE.__dict__['track2groups'].keys())
replicates = [str(x).split("-")[2] for x in rep_agg]
time_rep_comb = [x for x in itertools.product(time_points, replicates)]
time_cond = ro.StrVector([x[0] for x in time_rep_comb])
rep_cond = ro.StrVector([x[1] for x in time_rep_comb])
ref_gtf = str(infile).split("-")[1]
condition = (str(infile).split("-")[0]).strip("deseq.dir/")
time_points = ",".join([str(i) for i in time_points])
replicates = ",".join(replicates)
statement = '''
cgat data2resamples
--log=%(outfile)s.log
--time=%(time_points)s
--replicates=%(replicates)s
--condition=%(condition)s
--resamples=%(resampling_resample)s
--input-gtf=%(ref_gtf)s
--output-file-directory=clustering.dir
--seed=%(resampling_seed)s
%(infile)s
'''
P.run()
P.touch(outfile)
def computeGOSimilarity(self, funcData):
print(funcData)
root = etree.fromstring(funcData)
rows = root.xpath('//row')
drugs=[]
drugsToGo={}
for row in rows:
d = row.xpath('./drug/text()')[0].strip()
s = row.xpath('./GO/text()')[0].strip()
if d in drugsToGo:
l = drugsToGo[d]
else:
l = []
drugsToGo[d] = l
l.append(s)
result = '<?xml version="1.0"?>'
result += '<data>'
compute = robjects.r['mgoSim']
for d1 in drugsToGo.keys():
for d2 in drugsToGo.keys():
print(drugsToGo[d1])
print(drugsToGo[d2])
x = compute(robjects.StrVector(drugsToGo[d1]), robjects.StrVector(drugsToGo[d2]), ont='MF', organism="human", measure="Wang")
result += "<row>"
result += "<drug1>" + d1 + "</drug1>"
result += "<drug2>" + d2 + "</drug2>"
result += "<sim>" + str(x[0]) + "</sim>"
result += "</row>"
result += "</data>"
return result
def calculate_degree_modularity(self, targetModule):
'''
calculates in degree (kIn) and out degree (kOut)
for the target module
'''
members = self.__get_module_members(targetModule)
degree = rsnippets.degree(self.adjacency, ro.StrVector(members),
self.args.edgeWeight)
self.modules[targetModule]['kIn'] = int(degree.rx2('kIn')[0])
self.modules[targetModule]['kOut'] = int(degree.rx2('kOut')[0])
size = self.modules[targetModule]['size']
self.modules[targetModule]['density'] = float(
self.modules[targetModule]['kIn']) / (float(size) *
(float(size) - 1.0) / 2.0)
def build_dataframe(monitor: Monitor) -> dplyr.DataFrame:
what = (
'susceptible',
'incubating',
'sick',
)
dataf = dplyr.DataFrame({
'what':
ro.StrVector([v for v in what for x in monitor.day]),
'day':
ro.IntVector([v for x in what for v in monitor.day]),
'count':
ro.IntVector([v for x in what for v in getattr(monitor, x)])
})
return dataf
def Morris(repet, factors, binf, bsup):
""" Simplified import of R'Morris function"""
factors = robj.StrVector(factors)
binf = numpy.array(binf)
bsup = numpy.array(bsup)
d = r.list('oat', 5, 3)
d = r['names<-'](d, ['type', 'levels', 'grid.jump'])
m = r.morris(factors=factors, r=repet, design=d, binf=binf, bsup=bsup)
param = r['data.frame'](m.r["X"][0])
pdict = dict(
(k, numpy.array(param.r[k][0]).tolist()) for k in r.colnames(param))
l = len(pdict[pdict.keys()[0]])
print 'Computational cost Morris SA is %d' % l
return m, pdict
def run(self):
from rpy2 import robjects
VEGAN_SCRIPT = 'L:/resources/code/models/pre_process/gnn_vegan.r'
# Source the gnn_vegan R file
robjects.r.source(VEGAN_SCRIPT)
# Create an R vector to pass
var_vector = robjects.StrVector(self.variables)
# Create the vegan file
robjects.r.write_vegan(self.method, self.spp_file, self.env_file,
var_vector, self.id_field,
self.species_transform,
self.species_downweighting, self.ord_file)
def barPlot(dict_,
keysInOrder=None,
printCounts=True,
ylim=None,
*args,
**kwdargs):
""" Plot a bar plot
Args:
dict_: a dictionary of name -> value, where value is the height of the bar
use a collections.OrderedDict() to easily convey the order of the groups
keysInOrder: an optional ordering of the keys in dict_ (alternate option to using collections.OrderedDict)
printCounts: option to print the counts on top of each bar
additional kwdargs are passed directly to r.barplot()
"""
if not keysInOrder:
keysInOrder = dict_.keys()
heights = ro.FloatVector([dict_[key] for key in keysInOrder])
kwdargs["names.arg"] = ro.StrVector(keysInOrder)
if ylim is None:
if printCounts:
ylim = [min(heights), max(heights) * 1.1]
else:
ylim = [min(heights), max(heights)]
x = r.barplot(heights, ylim=ro.FloatVector(ylim), *args, **kwdargs)
if printCounts:
heightsStrings = ["{:.2g}".format(height) for height in heights]
r.text(x, ro.FloatVector(heights), ro.StrVector(heightsStrings), pos=3)
return x
def r_ttest(py_data1, py_data2, mu=0, alt="two.sided"):
"""
Defined for use within apply_ttest function.
"""
# Make sure datasets have equal dimensions
py_data1, py_data2 = equal_data(py_data1, py_data2)
# Bring in t.test function from R
t_test = R.r['t.test']
# Convert numpy array to R vector for both datasets
data1 = R.FloatVector(py_data1)
data2 = R.FloatVector(py_data2)
# Perform t-test on two populations
test = t_test(data1, data2, mu=mu,
**{'paired':True, 'na.action':R.StrVector(("na.exclude",)),
'alternative':R.StrVector((alt,))})
# Index names from R list and report the p-value
names = test.names
pval = test[names.index('p.value')][0]
return pval
def deseq_results(dds, condition1, condition2, out_dir):
#Get DESeq2 results.
to_dataframe = robjects.r('function(x) data.frame(x)')
res = to_dataframe(
deseq.results(dds,
contrast=robjects.StrVector(
['Group', condition1, condition2])))
gene_ids = res.rownames
res = pandas2ri.ri2py_dataframe(res)
res.index = gene_ids
#Output results.
res.to_csv(os.path.join(out_dir, f'{condition1}-{condition2}-results.txt'),
sep='\t')
def Morris(repeat, factors, binf, bsup):
""" Simplified import of R'Morris function"""
factors = robj.StrVector(factors)
binf = numpy.array(binf)
bsup = numpy.array(bsup)
d = r.list('oat', 5, 3)
d = r['names<-'](d, ['type', 'levels', 'grid.jump'])
m = r.morris(factors=factors, r=repeat, design=d, binf=binf, bsup=bsup)
#param=r['data.frame'](m.rx["X"])
param = r['data.frame'](r['$'](m, 'X'))
pdict = dict(
(k, numpy.array((r['$'](param, k))).tolist()) for k in param.names)
#pdict = dict((str(k), list(v)) for k,v in param.iteritems())
return m, pdict
def train_elastic_net_wrapper(features_data_,
features_,
d_,
data_annotation_,
x_w=None,
prune=True,
nested_folds=10):
x = numpy.array([features_data_[v] for v in features_.id.values])
dimnames = robjects.ListVector([(1, robjects.StrVector(d_["individual"])),
(2,
robjects.StrVector(features_.id.values))])
x = robjects.r["matrix"](robjects.FloatVector(x.flatten()),
ncol=features_.shape[0],
dimnames=dimnames)
y = robjects.FloatVector(d_[data_annotation_.gene_id])
nested_folds = robjects.FloatVector([nested_folds])
#py2ri chokes on None.
if x_w is None:
res = train_elastic_net(y, x, n_train_test_folds=nested_folds)
else:
res = train_elastic_net(
y, x, penalty_factor=x_w, n_train_test_folds=nested_folds
) # observation weights, not explanatory variable weight :( , x_weight = x_w)
return pandas2ri.ri2py(res[0]), pandas2ri.ri2py(res[1])
def selectRiverBarPlots(self, filename):
riverDict = {}
vectorDict = {}
if self.selectParm != None:
for i in self.selectParm:
riverDict = self.countAllRiverParm(i)
vectorDict['River'] = robjects.StrVector(riverDict.keys())
vectorDict['Count'] = robjects.IntVector(riverDict.values())
newFilename = filename + "_" + i + "_bar_river"
self.barPlot(robjects.DataFrame(vectorDict), newFilename,
"River", "Count")
def _generate_table(self, declarations):
"""Generates an R data frame table from the list of declarations."""
decl_table = defaultdict(list)
for decl in declarations:
decl_dict = self._map_fields(decl)
# R DataFrame is column-major.
for k, v in decl_dict.items():
decl_table[k].append(v)
return robjects.DataFrame(
# Have to translate into a properly typed vector, otherwise R will treat the data in a bad way.
{
k: (robjects.StrVector(v)
if k in STR_COLUMNS else robjects.FloatVector(v))
for k, v in decl_table.items()
})
def convert_to_r_dataframe(self, df, strings_as_factors=False):
"""
Convert a pandas DataFrame to a R data.frame.
Parameters
----------
df: The DataFrame being converted
strings_as_factors: Whether to turn strings into R factors (default: False)
Returns
-------
A R data.frame
"""
import rpy2.rlike.container as rlc
columns = rlc.OrdDict()
# FIXME: This doesn't handle MultiIndex
for column in df:
value = df[column]
value_type = value.dtype.type
if value_type == np.datetime64:
value = com.convert_to_r_posixct(value)
else:
value = [
item if pd.notnull(item) else com.NA_TYPES[value_type]
for item in value
]
value = com.VECTOR_TYPES[value_type](value)
if not strings_as_factors:
I = ro.baseenv.get("I")
value = I(value)
columns[column] = value
r_dataframe = ro.DataFrame(columns)
del columns
r_dataframe.rownames = ro.StrVector(list(df.index))
r_dataframe.colnames = list(df.columns)
return r_dataframe
def asVector(val: Iterable,
names: Optional[Iterable] = None) -> robj.Vector:
val = np.asarray(ll(val))
vect = {
'i': robj.IntVector,
'u': robj.IntVector,
'f': robj.FloatVector,
'b': robj.BoolVector,
'S': robj.StrVector,
'U': robj.StrVector,
}.get(val.dtype.kind, lambda x: None)(val)
if missing(vect):
raise TypeError(f'unknown vector type [{val.dtype.kind}]')
if available(names):
vect.names = robj.StrVector(np.asarray(ll(names), dtype=str))
return vect
def r_matrix(np_matrix, col_names=None):
"""Convert a numpymatrix to R matrix. If no columns are provided
it will assign the following ['x_1', 'x_2',... ] as column names
"""
if np_matrix.ndim != 2:
msg = 'Input input dimension is %s and MUST be 2' % np_matrix.ndim
raise ValueError(msg)
n_row, n_col = np_matrix.shape
r_mat = robjects.r.matrix(np_matrix, nrow=n_row, ncol=n_col)
if col_names is None:
col_names = ['x_%s' % (i + 1) for i in range(n_col)]
r_mat.colnames = robjects.StrVector(col_names)
return r_mat
