def phenograph_clustering_freq(d,DTCRU,n_jobs=1):
nbrs = NearestNeighbors(n_neighbors=30, metric='precomputed',n_jobs=n_jobs).fit(d)
d_knn, idx = nbrs.kneighbors(d)
IDX, _, _ = phenograph_local.cluster(d=d_knn, idx=idx, n_jobs=1)
label_id = DTCRU.class_id
file_id = DTCRU.sample_id
freq = DTCRU.freq
file_list = DTCRU.sample_list
DFs = []
DF_Sum = pd.DataFrame()
DF_Sum['File'] = file_list
DF_Sum.set_index('File', inplace=True)
for i in np.unique(IDX):
if i != -1:
sel = IDX == i
label_sel = label_id[sel]
file_sel = file_id[sel]
freq_sel = freq[sel]
df = pd.DataFrame()
df['Labels'] = label_sel
df['File'] = file_sel
df['Frequency'] = freq_sel
df_sum = df.groupby(by='File', sort=False).agg({'Frequency': 'sum'})
DF_Sum['Cluster_' + str(i)] = df_sum
DFs.append(df)
DF_Sum.fillna(0.0, inplace=True)
return DF_Sum, IDX
def phenograph_clustering(d):
nbrs = NearestNeighbors(n_neighbors=30, metric='precomputed').fit(d)
d, idx = nbrs.kneighbors(d)
IDX, _, _ = phenograph_local.cluster(d=d, idx=idx, n_jobs=1)
c_freq = []
for i in np.unique(IDX):
if i != -1:
sel = IDX == i
c_freq.append(np.sum(sel))
return c_freq