
import pandas as pd
import numpy as np
import scanpy as sc
from NSForest_v3dot9_1 import *
import plotly.express as px
import nsforest.nstest as ns
import matplotlib.pyplot as plt
from math import log


data_folder = "demo data/"
pd.set_option('display.max_rows', None)

adata = sc.read_h5ad(data_folder + "adata_MTG.h5ad") 
cluster_header = "cluster_updated"


df_cluster_size = pd.DataFrame({'cluster_size': adata.obs[cluster_header].value_counts()})
cluster_size = df_cluster_size.cluster_size


cells = list(adata.obs.index) #list of all cells
clusters = adata.obs[cluster_header].value_counts().index.tolist() #75 clusters
all_clusters=np.array(adata.obs.iloc[:,1]) #Repeated clusters for all cells

cells_cluster = list(zip(cells,all_clusters)) #matrix of only cells and clusters
df = pd.DataFrame(cells_cluster, columns=['cells','clusters'])

number_of_cells = [[None]]*len(clusters)
cells_list = [[None]]*len(clusters)
classification = np.column_stack((clusters, cells_list, number_of_cells))

for i in range(len(clusters)):
    classification[i][1]=np.array(df[df.clusters == clusters[i]]['cells'])
    classification[i][2]=len(classification[i][1])

df2 = pd.DataFrame(classification, columns=['Cluster', 'Cells','Number of Cells'])
df2 #list of clusters, cluster-specific cells, and number of cells


#Set x = threshold for cluster size. Select x number of cells from each 
#cluster. If cluster size < x, select all cells from that cluster. Try x = 50, 100, 
#200, 300, 500. Plot the F-beta vs cluster size figure for each threshold. 
#Repeat y times for each x. Use y = 5.
thresholds = [50,100,200,300,500]
df_fscores = pd.DataFrame()
for i in range(len(thresholds)): #repeating for each threshold
    adata = sc.read_h5ad(data_folder + "adata_MTG.h5ad") 
    final_cells = []
    cluster_size = []
    cells_cluster_df = pd.DataFrame(cells_cluster, columns=['cells','clusters'])
    for j in range(len(df2)): #repeating for every cluster 
        if thresholds[i] < df2.iloc[j,2]:
            randomly_selected_cells = np.random.choice(df2.iloc[j,1], size=thresholds[i], replace=False)
        else:
            randomly_selected_cells = df2.iloc[j,1]
        cluster_size.append(len(randomly_selected_cells)) #to make the list of cluster size for plotting
        final_cells.extend(randomly_selected_cells) 
    cells_cluster_df = cells_cluster_df[cells_cluster_df['cells'].isin(final_cells)]
    adata = adata[cells_cluster_df['cells']] 
    NSForest_results = NSForest(adata, cluster_header=cluster_header, n_trees=100, n_genes_eval=6) 
    #NSForest_results['cluster_size'] = [log(y,2) for y in cluster_size]
    df_fscores['threshold:'+str(thresholds[i])] = NSForest_results['f_score']
fig = df_fscores.boxplot(rot=45)
fig.set_ylabel('F-beta scores')
boxplot = fig.get_figure()
boxplot.savefig('boxplot_1.pdf', bbox_inches='tight')

	Cluster	Cells	Number of Cells
0	Exc L2-3 LINC00507 FREM3	[F1S4_160106_038_C01, F1S4_160106_038_E01, F1S...	2284
1	Exc L5-6 THEMIS C1QL3	[F1S4_160106_003_B01, F1S4_160106_003_D01, F1S...	1537
2	Exc L3-5 RORB ESR1	[F1S4_160106_001_H01, F1S4_160106_002_C01, F1S...	1428
3	Exc L4-5 RORB FOLH1B	[F1S4_160106_002_D01, F1S4_160106_002_H01, F1S...	870
4	Exc L2 LAMP5 LTK	[F1S4_160106_038_D01, F1S4_160106_039_B01, F1S...	812
5	Exc L4-6 RORB SEMA3E	[F1S4_160106_003_E01, F1S4_160106_009_G01, F1S...	777
6	Inh L2-4 PVALB WFDC2	[F1S4_160106_059_D01, F1S4_160106_060_F01, F1S...	387
7	Exc L5-6 FEZF2 ABO	[F1S4_160106_001_E01, F1S4_160106_002_F01, F1S...	373
8	Inh L1-4 LAMP5 LCP2	[F1S4_160106_002_G01, F1S4_160106_039_H01, F1S...	356
9	Exc L4-6 FEZF2 IL26	[F1S4_160106_006_C01, F1S4_160106_008_C01, F1S...	344
10	Exc L5-6 FEZF2 EFTUD1P1	[F1S4_160106_001_G01, F1S4_160317_001_F01, F1S...	314
11	Oligo L1-6 OPALIN	[F1S4_160106_003_A01, F1S4_160106_071_A01, F1S...	313
12	Inh L1 LAMP5 NMBR	[F1S4_160106_037_G01, F1S4_160106_040_E01, F1S...	283
13	Exc L3-4 RORB CARM1P1	[F1S4_160218_040_B01, F1S4_160218_044_D01, F1S...	280
14	Inh L1-3 SST CALB1	[F1S4_160317_003_C01, F1S4_160721_001_A01, F1S...	279
15	Inh L2-6 LAMP5 CA1	[F1S4_160106_048_D01, F1S4_160106_058_F01, F1S...	256
16	OPC L1-6 PDGFRA	[F1S4_160106_037_A01, F1S4_160106_041_A01, F1S...	238
17	Astro L1-6 FGFR3 SLC14A1	[F1S4_160106_040_A01, F1S4_160106_047_A01, F1S...	230
18	Inh L4-6 SST B3GAT2	[F1S4_160106_001_B01, F1S4_160218_025_A01, F1S...	182
19	Inh L1-3 VIP CHRM2	[F1S4_160106_043_E01, F1S4_160106_044_H01, F1S...	175
20	Exc L4-5 RORB DAPK2	[F1S4_160106_006_F01, F1S4_160218_026_B01, F1S...	173
21	Exc L2-4 LINC00507 GLP2R	[F1S4_160721_037_A01, F1S4_160721_037_B01, F1S...	170
22	Inh L4-6 PVALB SULF1	[F1S4_160106_004_C01, F1S4_160218_028_F01, F1S...	167
23	Exc L5-6 RORB TTC12	[F1S4_160106_001_C01, F1S4_160317_003_D01, F1S...	167
24	Exc L3-5 RORB COL22A1	[F1S4_160218_033_E01, F1S4_160218_038_C01, F1S...	160
25	Exc L4-6 RORB C1R	[F1S4_160317_003_G01, F1S4_160721_077_F01, F1S...	160
26	Exc L3-5 RORB FILIP1L	[F1S4_160218_038_B01, F1S4_160218_044_G01, F1S...	153
27	Exc L5-6 THEMIS CRABP1	[F1S4_160106_005_D01, F1S4_160721_102_D01, F1S...	147
28	Inh L3-5 SST ADGRG6	[F1S4_160218_034_E01, F1S4_160721_035_B01, F1S...	132
29	Inh L1-2 SST BAGE2	[F1S4_160106_039_D01, F1S4_160106_039_E01, F1S...	108
30	Inh L1-2 ADARB2 MC4R	[F1S4_160106_044_C01, F1S4_160106_045_C01, F1S...	107
31	Exc L3-5 RORB TWIST2	[F1S4_160106_006_G01, F1S4_160218_032_B01, F1S...	93
32	Inh L4-5 SST STK32A	[F1S4_160218_025_B01, F1S4_160218_032_F01, F1S...	93
33	Inh L1-2 PAX6 CDH12	[F1S4_160106_037_E01, F1S4_160106_045_G01, F1S...	90
34	Inh L3-6 VIP HS3ST3A1	[F1S4_160218_036_G01, F1S4_160526_030_B01, F1S...	80
35	Inh L5-6 SST MIR548F2	[F1S4_160106_009_A01, F1S4_160317_001_E01, F1S...	80
36	Inh L5-6 SST NPM1P10	[F1S4_160106_008_G01, F1S4_160218_035_C01, F1S...	79
37	Exc L5-6 THEMIS FGF10	[F1S4_160317_011_F01, F1S4_160721_087_D01, F1S...	78
38	Inh L1-3 VIP ADAMTSL1	[F1S4_160106_068_B01, F1S4_160106_069_H01, F1S...	72
39	Inh L1-3 VIP GGH	[F1S4_160106_048_A01, F1S4_160831_008_B01, F1S...	68
40	Inh L2-4 VIP CBLN1	[F1S4_160106_066_H01, F1S4_160721_001_B01, F1S...	67
41	Inh L4-5 PVALB MEPE	[F1S4_160106_009_B01, F1S4_160721_086_G01, F1S...	64
42	Inh L2-4 SST FRZB	[F1S4_160106_066_G01, F1S4_160218_039_B01, F1S...	64
43	Inh L1-3 VIP CCDC184	[F1S4_160106_059_G01, F1S4_160106_060_G01, F1S...	64
44	Inh L5-6 SST KLHDC8A	[F1S4_160317_003_B01, F1S4_160721_123_D01, F1S...	63
45	Micro L1-6 TYROBP	[F1S4_160106_045_A01, F1S4_160721_063_H01, F1S...	63
46	Inh L2-5 VIP TYR	[F1S4_160106_047_G01, F1S4_160218_035_G01, F1S...	62
47	Astro L1-2 FGFR3 GFAP	[F1S4_160106_038_A01, F1S4_160106_039_A01, F1S...	61
48	Inh L1-2 VIP PCDH20	[F1S4_160106_037_B01, F1S4_160106_044_D01, F1S...	61
49	Inh L3-6 SST HPGD	[F1S4_160106_007_E01, F1S4_160721_119_F01, F1S...	60
50	Exc L5-6 FEZF2 IL15	[F1S4_160721_112_E01, F1S4_160802_145_F01, F1S...	56
51	Inh L2-5 VIP SERPINF1	[F1S4_160218_034_C01, F1S4_160218_035_E01, F1S...	55
52	Exc L5-6 THEMIS DCSTAMP	[F1S4_160721_101_B01, F1S4_160721_117_B01, F1S...	53
53	Inh L1-4 VIP OPRM1	[F1S4_160106_054_A01, F1S4_160106_058_G01, F1S...	52
54	Exc L6 FEZF2 SCUBE1	[F1S4_160721_134_E01, F1S4_160802_147_C01, F1S...	52
55	Inh L5-6 PVALB LGR5	[F1S4_160721_107_F01, F1S4_160721_111_F01, F1S...	52
56	Inh L1 SST CHRNA4	[F1S4_160106_044_E01, F1S4_160106_047_F01, F1S...	52
57	Inh L1-2 VIP LBH	[F1S4_160106_060_D01, F1S4_160106_064_D01, F1S...	47
58	Inh L2-3 VIP CASC6	[F1S4_160106_003_C01, F1S4_160106_053_D01, F1S...	45
59	Inh L1-2 VIP TSPAN12	[F1S4_160106_040_C01, F1S4_160902_069_F01, F1S...	42
60	Inh L4-6 SST GXYLT2	[F1S4_160721_071_F01, F1S4_160721_094_B01, F1S...	41
61	Inh L2-6 VIP QPCT	[F1S4_160721_045_D01, F1S4_160721_087_E01, F1S...	37
62	Inh L2-4 VIP SPAG17	[F1S4_160721_048_G01, F1S4_160802_057_G01, F1S...	33
63	Inh L2-5 PVALB SCUBE3	[F1S4_160218_028_A01, F1S4_160721_083_H01, F1S...	32
64	Inh L1-3 VIP SYT6	[F1S4_160106_039_C01, F1S4_160106_043_G01, F1S...	29
65	Inh L5-6 SST TH	[F1S4_160106_009_C01, F1S4_160721_118_D01, F1S...	27
66	Inh L5-6 LHX6 GLP1R	[F1S4_160721_135_C01, F1S4_160802_122_F01, F1S...	27
67	Inh L1-4 VIP CHRNA6	[F1S4_160721_051_F01, F1S4_160831_050_G01, F1S...	25
68	Exc L4-5 FEZF2 SCN4B	[F1S4_160526_030_F01, F1S4_160831_081_D01, F1S...	25
69	Inh L1-2 LAMP5 DBP	[F1S4_161026_002_D01, F1S4_170106_139_A01, F1S...	21
70	Exc L6 FEZF2 OR2T8	[F1S4_161026_151_F01, F1S4_161102_146_A01, F1S...	19
71	Inh L1-4 VIP PENK	[F1S4_160831_058_B01, F1S4_160831_084_B01, F1S...	17
72	Inh L1-2 PAX6 TNFAIP8L3	[F1S4_170106_013_G01, F1S4_170106_127_B01, F1S...	16
73	Inh L3-6 SST NPY	[F1S4_160721_101_E01, F1S4_160721_111_D01, F1S...	15
74	Endo L2-6 NOSTRIN	[F1S4_160218_030_H01, F1S4_160721_059_H01, F1S...	9

Data¶

Cluster Size¶

Cluster-specific cells¶

Task 2. F-beta vs. cluster size for each threshold¶