def _compute_adjacency(self):
info = self.pairwise_information_
method = self.adjacency_method
kwargs = self.adjacency_method_kwargs
if kwargs is None:
kwargs = {}
adjacency_matrix = None
if method == 'raw':
adjacency_matrix = info
else:
minet = importr('minet')
if method == 'mrnet':
func = minet.mrnet
elif method == 'clr':
func = minet.clr
elif method == 'aracne':
func = minet.aracne
else:
raise ValueError('Unsupported method {!r}'.format(method))
ans = func(numpy2ri.numpy2rpy(np.asarray(info)), **kwargs)
ans = np.asarray(ans)
ans = pd.DataFrame(ans, index=info.index, columns=info.columns)
adjacency_matrix = ans
self.adjacency_matrix_ = adjacency_matrix
self.adjacency_ = stack_triu(adjacency_matrix)
def py2rpy_pandasindex(obj):
if obj.dtype.kind == 'O':
return StrVector(obj)
else:
# pandas2ri should definitely not have to know which paths remain to be
# converted by numpy2ri
# Answer: the thing is that pandas2ri builds on the conversion
# rules defined by numpy2ri - deferring to numpy2ri is allowing
# us to reuse that code.
return numpy2ri.numpy2rpy(obj)
from rpy2.robjects.packages import importr
import numpy as np
json_f = sys.argv[1]
brain_part = sys.argv[2]
with open(json_f, "r") as jf:
ce_dict = json.load(jf)
for region in ce_dict:
for stage in ce_dict[region]:
ce_wdict = []
for c in ce_dict[region][stage]:
cluster = ce_dict[region][stage][c]
mat = np.array([[cluster["x"], cluster["m_x"]],
[cluster["T_x"], cluster["n"]]])
r_cont = numpy2rpy(mat)
r.assign("cont", r_cont)
r('fmat_pval <- fisher.test(cont, alternative="greater",conf.int=TRUE)$p.value'
)
r('fmat_estimate <- fisher.test(cont, alternative="greater",conf.int=TRUE)$estimate'
)
r('fmat_conf <- fisher.test(cont, alternative="greater",conf.int=TRUE)$conf.int'
)
pval = r("fmat_pval")
odds = r("fmat_estimate")
conf_int = r("fmat_conf")
ce_wdict.append({
"Cluster_name":
cluster["Cluster_num"],
"CE_Stage_region":
cluster["x"],
jdict[fstr]=json.load(wr)
comparisons=["prenatal_adulthood",
"early_childhood_adulthood"]
i=1
for cnum in range(0,cluster_num+1):
region_d={}
for r in regions:
comp_d={}
for comparison in comparisons:
jd="proportion"+"_"+r+"_"+comparison
clus_details=jdict[jd]["Cluster"+str(cnum)]
ce=np.array([clus_details["a"],clus_details["b"]])
univ=np.array([clus_details["U"],clus_details["U"]])
r_ce=numpy2rpy(ce)
r_univ=numpy2rpy(univ)
R.assign("ce", r_ce)
R.assign("univ", r_univ)
omat=FloatVector([clus_details["a"],clus_details["b"],clus_details["U"],clus_details["U"]])
R('pmat <- prop.test(x=ce,n=univ,alternative="greater",conf.level=0.95)$p.value')
pval=float((str(R("pmat")).split(" ")[1]).strip("\n"))
epiod=epitools.oddsratio(omat)
odds=epiod[1][1]
comp_d[comparison+"_pval"]=pval
comp_d[comparison+"_pval_adj"]=float(pval)*float(2)*(float(cluster_num+1))
comp_d[comparison+"_odds_ratio"]=odds
region_d[r]=comp_d
with open(os.path.join(work_dir,brain_part+"_prop_summary",brain_part+"_proportion_cluster"+str(cnum)+".csv"),"w") as wh:
writer=csv.writer(wh)