from sklearn.cluster import AffinityPropagation
import matplotlib.pyplot as plt
import pandas as pd
import numpy as np
import seaborn as sns
%matplotlib inline
from itertools import cycle

data = pd.read_csv('../datareader/score/old-score.csv')
data.index = data['phrase']
ndf = data.drop('phrase',axis =1)

ndf.head()

ndf.shape

(3175, 7)

tdata = ndf.copy(deep =True)

cva_ihd = tdata.drop([ 'CM', 'ARR', 'VD', 'CHD'],axis =1)
cva_cm = tdata.drop([ 'IHD', 'ARR', 'VD', 'CHD'],axis =1)
cva_arr = tdata.drop([ 'IHD', 'CM', 'VD', 'CHD'],axis =1)
cva_vd = tdata.drop([ 'IHD', 'CM', 'ARR', 'CHD'],axis =1)
cva_chd = tdata.drop(['IHD', 'CM', 'ARR', 'VD'],axis =1)
ihd_cm = tdata.drop(['CVA', 'ARR', 'VD', 'CHD'],axis =1)
ihd_arr = tdata.drop(['CVA', 'CM', 'VD', 'CHD'],axis =1)
ihd_vd = tdata.drop(['CVA', 'CM', 'ARR', 'CHD'],axis =1)
ihd_chd = tdata.drop(['CVA', 'CM', 'ARR', 'VD'],axis =1)
cm_arr = tdata.drop(['CVA', 'IHD', 'VD', 'CHD'],axis =1)
cm_vd = tdata.drop(['CVA', 'IHD', 'ARR', 'CHD'],axis =1)
cm_chd = tdata.drop(['CVA', 'IHD', 'ARR', 'VD'],axis =1)
arr_vd = tdata.drop(['CVA', 'IHD', 'CM', 'CHD'],axis =1)
arr_chd = tdata.drop(['CVA', 'IHD', 'CM', 'VD'],axis =1)
vd_chd = tdata.drop(['CVA', 'IHD', 'CM', 'ARR'],axis =1)

def remover(df,cvd1,cvd2):

    protein = df.index
    d1 = df[cvd1]
    d2 = df[cvd2]
    DT = []
    c = 0.5
    cutoff = list(df.mean()+(c*df.std()))
    for p,a,b in zip(protein,d1,d2):
        if a > cutoff[0] or  b > cutoff[1] :
            DT.append({'protein': p, cvd1: a, cvd2: b},)
    dfr = pd.DataFrame(DT)
    dfr.index = dfr['protein']
    dfr = dfr.drop('protein',axis =1)
    return dfr

cvd1 = 'ARR' #
cvd2 = 'CVA'
df = cva_arr
pref=-1.0

cvd1 = 'IHD' #
cvd2 = 'CVA'
df = cva_ihd
pref=-1.0

cvd1 = 'CM' #
cvd2 = 'CVA'
df = cva_cm
pref=-0.1

cvd1 = 'VD'  #
cvd2 = 'CVA'
df = cva_vd
pref=-0.1

cvd1 = 'CHD' #
cvd2 = 'CVA'
df = cva_chd
pref=-0.1

cvd1 = 'IHD'#
cvd2 = 'CM'
df = ihd_cm
pref=-0.1

cvd1 = 'IHD'#
cvd2 = 'ARR'
df = ihd_arr
pref=-0.1

cvd1 = 'IHD'#
cvd2 = 'VD'
df = ihd_vd
pref=-0.1

cvd1 = 'IHD' #
cvd2 = 'CHD'
df = ihd_chd
pref=-0.1

cvd1 = 'CM' #
cvd2 = 'ARR'
df = cm_arr
pref=-0.1

cvd1 = 'CM' #
cvd2 = 'VD'
df = cm_vd
pref=-0.1

cvd1 = 'CM'#
cvd2 = 'CHD'
df = cm_chd
pref=-0.1

cvd1 = 'ARR'#
cvd2 = 'VD'
df = arr_vd
pref=-0.1

cvd1 = 'ARR'#
cvd2 = 'CHD'
df = arr_chd
pref=-0.1

cvd1 = 'VD'#
cvd2 = 'CHD'
df = vd_chd
pref=-0.1

rdf = remover(df,cvd1,cvd2)
rdf.head()

rdf.shape

(628, 2)

X = rdf.copy(deep =True)
p = np.array(list(X.index))

# Compute Affinity Propagation
af = AffinityPropagation(preference = pref).fit(X)
indices = af.cluster_centers_indices_
labels = af.labels_
n_clusters = len(indices)
n_clusters

23

sns.set()
plt.figure(figsize = [15,15])
DTT =[]
X = np.array(X)
colors = cycle('bgrcmykbgrcmykbgrcmykbgrcmyk')
for k, col in zip(range(n_clusters), colors):
    class_members = labels == k
    cluster_center = X[indices[k]]
    ln = len(X[class_members, 0])
    if ln >1:
        for i in range(ln):
            DTT.append({'label': k,\
                        'protein': p[class_members][i],\
                        cvd1: X[class_members, 0][i],\
                        cvd2: X[class_members, 1][i],\
                        'color':col})
    else:

        DTT.append({'label': k,\
                        'protein': p[class_members][0],\
                        cvd1: X[class_members, 0][0],\
                        cvd2: X[class_members, 1][0],\
                        'color':col})

    # plot data
    plt.plot(X[class_members, 0], X[class_members, 1], col + '.')
    # plot center
    plt.plot(cluster_center[0], cluster_center[1], '.', markerfacecolor=col,
             markeredgecolor='k', markersize=14)
    # draw lines
    for x in X[class_members]:
        plt.plot([cluster_center[0], x[0]], [cluster_center[1], x[1]], col)

plt.title('Estimated number of clusters: %d' % n_clusters)
plt.xlabel(cvd1, fontsize = 25)
plt.ylabel(cvd2, fontsize = 25 )
plt.savefig('plot/'+cvd1+'-'+cvd2+'.pdf')
plt.show()

listdf = pd.DataFrame(DTT)
listdf.index = listdf['protein']
listdf = listdf.drop('protein',axis =1)
listdf.head(5)

data = pd.read_csv('uniprot.csv')
data.index = data['protein']
udf = data.drop('protein',axis =1)
udf.head()

idx_cvd = list(listdf.index)
idx_uprt =  list(udf.index)

prot_dict = []
for item in idx_cvd:
    data = listdf.loc[item,:]
    if item in idx_uprt:
         prot_dict.append({'Protein':item,\
                     cvd1:data[0],\
                     cvd2:data[1],\
                     'color':data[2],\
                     'label': data[3],\
                     'uprot':udf.loc[item,:][0]})

    else:
        print(item,'Match not Found')
        prot_dict.append({'Protein':item,\
                     cvd1:data[0],\
                     cvd2:data[1],\
                     'color':data[2],\
                     'label': data[3],\
                     'uprot':'NAN'})

resultdf = pd.DataFrame(prot_dict)
resultdf.index = resultdf['Protein']
resultdf = resultdf.drop('Protein',axis =1)
resultdf.head(10)

resultdf.to_csv('data/' + cvd1 + '-' + cvd2 + '.csv')