complex network (gene Suc TF)

import matplotlib.pyplot as plt
import networkx as nx
import pandas as pd
import numpy as np
import matplotlib as mpl
mpl.rcParams['pdf.fonttype']=42
mpl.rcParams['ps.fonttype']=42

/Users/yuanzan/anaconda3/lib/python3.8/site-packages/pandas/core/computation/expressions.py:20: UserWarning: Pandas requires version '2.7.3' or newer of 'numexpr' (version '2.7.1' currently installed).
  from pandas.core.computation.check import NUMEXPR_INSTALLED

network data prepare

c1 = pd.read_excel("gene-TF相关性.xlsx", sheet_name=0, index_col=0)[['r', 'ID']]
c2 = pd.read_excel("gene-TF相关性.xlsx", sheet_name=1, index_col=0)[['r', 'ID']]
c3 = pd.read_excel("gene-TF相关性.xlsx", sheet_name=2, index_col=0)[['r', 'ID']]

def make_graD(df):
    df.columns = ['corr', 'ID']
    df['gene'] = df.index
    df['absR'] = abs(df['corr'])
    df['Sugar'] = 'Suc'
    df['edgeType'] = '-'
    df.loc[df.query('absR<0.6').index, 'edgeType'] = '-.-'
    df['Rlabel'] = round(df['corr'], 2)
    df['gene_r'] = df['gene'] + '\n' + df['Rlabel'].astype(str)
    df.index = df['ID']

    return df

c1 = make_graD(c1)
c2 = make_graD(c2)
c3 = make_graD(c3)




## gene TF corr
gene_TF_corr = pd.read_csv("TF_gene_FPKM_corr.csv", sep="\t")
gene_TF_corr['Gene_symbol'] = c1.loc[gene_TF_corr['Gene'], "gene"].to_list()
gene_TF_corr['TF_symbol'] = pd.concat([c2, c3], axis=0).loc[gene_TF_corr['TF'], "gene"].to_list()

gene_TF_corr['absR'] = abs(gene_TF_corr['corr'])
gene_TF_corr['edgeType'] = '-'
gene_TF_corr.loc[gene_TF_corr.query('absR<0.6').index, 'edgeType'] = '-.-'
gene_TF_corr['Rlabel'] = round(gene_TF_corr['corr'], 2)
#gene_TF_corr['gene_r'] = gene_TF_corr['gene'] + '\n' + gene_TF_corr['Rlabel'].astype(str)

cols = ['#FBE28F', '#A3B6E7',
        '#B0B3B7', '#DAC7AE',
        '#6CB0D2', '#E3ECC2',
        '#7FA6CD', '#E59386',
        '#8DA07E', '#7B687E',
        '#738E89' , '#A97D75',
        '#85B8B4' , '#F2B0D2']

TF_family = ['ERF', 'MYB', 'NAC', 'WRKY', 'B3', 'HLH', 'HSF', 'TCP', 'Trihelix', 'VOZ', 'ZIP']
TF_cols = pd.DataFrame(TF_family)
TF_cols.columns = ['TFFs']
TF_cols.index = TF_cols['TFFs']
TF_cols['color'] = None

for i,v in enumerate(TF_family):
    TF_cols.loc[v, "color"] = cols[i]

TF_cols.index.name = 'ID'
TF_cols = TF_cols.sort_values(['TFFs'])

network style 1

def get_nodes(df, scale=1):
    G = nx.cycle_graph(df.shape[0])
    pos = nx.circular_layout(G, scale=scale)
    pos = {df['gene'].to_list()[i]:pos[i] for i in pos}

    return G, pos

plt.figure(1, figsize=(10, 10))

mpl.rcParams['pdf.fonttype']=42
mpl.rcParams['ps.fonttype']=42


G1, pos1 = get_nodes(c1.sort_values(['corr']), scale=7)

c23 = pd.concat([c2, c3], axis=0)


c23['TFfamily'] = None

for i,row in TF_cols.iterrows():
    c23.loc[c23[c23['gene'].str.contains(row['TFFs'])].index, "TFfamily"] = row['TFFs']

c23['TFcolor'] = TF_cols.loc[c23['TFfamily'], 'color'].to_list()
c23 = c23.sort_values(['corr', 'TFfamily'])




G23, pos23 = get_nodes(c23, scale=10)




G = nx.Graph()
G.clear()

for i in c1['gene'].to_list():
    G.add_node(i)

for i in c23['gene'].to_list():
    G.add_node(i, )

G.add_node("Suc")

pos = dict(pos1, **pos23)
pos['Suc'] = np.array([0,0])





nx.draw_networkx_nodes(G, {'Suc': pos['Suc']}, nodelist=['Suc'], node_size=5000, node_color='#C7442A')
nx.draw_networkx_nodes(G, pos1, nodelist=c1['gene'].to_list(), node_size=300, node_color='#E0CC71')


for i in c23['TFfamily'].unique():
    nx.draw_networkx_nodes(G, pos23, nodelist=c23[c23['TFfamily']==i]['gene'].to_list(), node_size=200, node_color=TF_cols.loc[i, 'color'])


nx.draw_networkx_labels(G, {'Suc': pos['Suc']}, labels={'Suc': 'Suc'}, font_size=12, font_family="sans-serif", font_color='w')


nx.draw_networkx_labels(G, pos1, labels={i:i for i in pos1}, font_size=10, font_family="sans-serif", font_color='k')
nx.draw_networkx_labels(G, pos23, labels={i:i for i in pos23}, font_size=5, font_family="sans-serif", font_color='k')




def draw_edge(df, G, pos, edge_color='#E00671', width=5, alpha=1):
    # add edges
    #c_edges = []
    c_edges_pos = []
    c_edges_neg = []
    for i,row in df.iterrows():
        G.add_edge(row['gene'], row['Sugar'], length=row['absR'], weight=row['corr'])
        #c_edges.append((row['gene'], row['Sugar']))
        if row['corr'] > 0:
            c_edges_pos.append((row['gene'], row['Sugar']))
        else:
            c_edges_neg.append((row['gene'], row['Sugar']))


    nx.draw_networkx_edges(G, pos, edgelist=c_edges_pos, edge_color=edge_color, width=width, alpha=alpha, style = 'solid')
    nx.draw_networkx_edges(G, pos, edgelist=c_edges_neg, edge_color=edge_color, width=width*1.2, alpha=alpha, style = 'dashed')

draw_edge(c1, G, pos, edge_color='#E00671', width=2)
draw_edge(c23, G, pos, edge_color='#07CCD2', width=1, alpha=0.5)


#gene_TF_edges = []
gene_TF_edges_pos = []
gene_TF_edges_neg = []

for i,row in gene_TF_corr.query('absR>0.6').iterrows():
    G.add_edge(row['Gene_symbol'], row['TF_symbol'], length=row['absR'], weight=row['corr'])
    #gene_TF_edges.append((row['Gene_symbol'], row['TF_symbol']))
    if row['corr'] > 0:
        gene_TF_edges_pos.append((row['Gene_symbol'], row['TF_symbol']))
    else:
        gene_TF_edges_neg.append((row['Gene_symbol'], row['TF_symbol']))


nx.draw_networkx_edges(G, pos, edgelist=gene_TF_edges_pos, edge_color='#E0CCD2',  width=0.6, alpha=1, )
nx.draw_networkx_edges(G, pos, edgelist=gene_TF_edges_neg, edge_color='#E0CCD2',  width=1, alpha=1, style="dashed")



ax = plt.gca()

for i, row in TF_cols.iterrows():
    ax.scatter(None,None, label=row['TFFs'], color=row['color'])

ax.legend(loc=5)

ax.margins(0.15)
plt.axis("off")
plt.tight_layout()


plt.savefig("multi_circle_network_style1.pdf")

network style2

import bezier
from matplotlib.collections import LineCollection
import random

def curved_graph(graph, pos = None, eps=0.2, pointn=30):

    if pos == None:
        pos = nx.spring_layout(graph)

    for u,v in graph.edges():
        x0, y0 = pos[u]
        x1, y1 = pos[v]
        segs = curved_line(x0,y0,x1,y1)
        lc = LineCollection(segs)
        plt.gca().add_collection(lc)
        plt.gca().autoscale_view()

def curved_line(x0, y0, x1, y1, eps=0.2, pointn=30):

    x2 = (x0+x1)/2.0 + 0.1 ** (eps+abs(x0-x1)) * (-1) ** (random.randint(1,4))
    y2 = (y0+y1)/2.0 + 0.1 ** (eps+abs(y0-y1)) * (-1) ** (random.randint(1,4))
    nodes = np.asfortranarray([
        [x0, x2, x1],
        [y0, y2, y1]
    ])
    curve = bezier.Curve(nodes,
                         degree=2)
    s_vals = np.linspace(0.0, 1.0, pointn)
    data=curve.evaluate_multi(s_vals)
    x=data[0]
    y=data[1]
    segments =[]
    for index in range(0,len(x)):
        segments.append([x[index],y[index]])
    segments = [segments]
    return  segments

def get_nodes(df, scale=1, center=np.array([0,0])):
    G = nx.cycle_graph(df.shape[0])
    pos = nx.circular_layout(G, scale=scale, center=center)
    pos = {df['gene'].to_list()[i]:pos[i] for i in pos}

    return G, pos

plt.figure(1, figsize=(12, 6))



c23 = pd.concat([c2, c3], axis=0)


c23['TFfamily'] = None

for i,row in TF_cols.iterrows():
    c23.loc[c23[c23['gene'].str.contains(row['TFFs'])].index, "TFfamily"] = row['TFFs']

c23['TFcolor'] = TF_cols.loc[c23['TFfamily'], 'color'].to_list()
c23 = c23.sort_values(['corr', 'TFfamily'])



G1, pos1 = get_nodes(c1.sort_values(['corr']), scale=2, center=np.array([-10,0]))
G23, pos23 = get_nodes(c23, scale=10, center=np.array([20,0]))




G = nx.Graph()
G.clear()

for i in c1['gene'].to_list():
    G.add_node(i)

for i in c23['gene'].to_list():
    G.add_node(i, )

G.add_node("Suc")


pos1_y = [9,7,5,3,1,-1,-3,-5,-7,-9]
pos1 = {v:np.array([-15, pos1_y[i]]) for i,v in enumerate(c1.sort_values(['corr'])['gene'].to_list())}

pos = dict(pos1, **pos23)
pos['Suc'] = np.array([0,0])



nx.draw_networkx_nodes(G, {'Suc': pos['Suc']}, nodelist=['Suc'], node_size=500, node_color='#C7442A')
nx.draw_networkx_nodes(G, pos1, nodelist=c1['gene'].to_list(), node_size=300, node_color='#E0CC71')


for i in c23['TFfamily'].unique():
    nx.draw_networkx_nodes(G, pos23, nodelist=c23[c23['TFfamily']==i]['gene'].to_list(), node_size=50, node_color=TF_cols.loc[i, 'color'])


nx.draw_networkx_labels(G, {'Suc': pos['Suc']}, labels={'Suc': 'Suc'}, font_size=12, font_family="sans-serif", font_color='w')


nx.draw_networkx_labels(G, pos1, labels={i:i for i in pos1}, font_size=10, font_family="sans-serif", font_color='k')
nx.draw_networkx_labels(G, pos23, labels={i:i for i in pos23}, font_size=6, font_family="sans-serif", font_color='k')




def draw_edge(df, G, pos, pos_edge_color='#E00671', neg_edge_color='#E00671', width=5, alpha=1):
    # add edges
    #c_edges = []
    c_edges_pos = []
    c_edges_neg = []
    for i,row in df.iterrows():
        G.add_edge(row['gene'], row['Sugar'], length=row['absR'], weight=row['corr'])
        #c_edges.append((row['gene'], row['Sugar']))
        if row['corr'] > 0:
            c_edges_pos.append((row['gene'], row['Sugar']))
        else:
            c_edges_neg.append((row['gene'], row['Sugar']))


    nx.draw_networkx_edges(G, pos, edgelist=c_edges_pos, edge_color=pos_edge_color, width=width*1.2, alpha=1, style = 'solid')
    nx.draw_networkx_edges(G, pos, edgelist=c_edges_neg, edge_color=neg_edge_color, width=width, alpha=0.6, style = 'dashed')

draw_edge(c1, G, pos, pos_edge_color='#E00671', neg_edge_color='g', width=2)
draw_edge(c23, G, pos, pos_edge_color='#E0CCD2', neg_edge_color='g', width=1, alpha=0.5)


#gene_TF_edges = []
gene_TF_edges_pos = []
gene_TF_edges_neg = []

for i,row in gene_TF_corr.query('absR>0.6').iterrows():
    G.add_edge(row['Gene_symbol'], row['TF_symbol'], length=row['absR'], weight=row['corr'])
    #gene_TF_edges.append((row['Gene_symbol'], row['TF_symbol']))
    if row['corr'] > 0:
        gene_TF_edges_pos.append((row['Gene_symbol'], row['TF_symbol']))
    else:
        gene_TF_edges_neg.append((row['Gene_symbol'], row['TF_symbol']))


nx.draw_networkx_edges(G, pos, edgelist=gene_TF_edges_pos, edge_color='#A0CBE2',  width=1, alpha=0.5 )
nx.draw_networkx_edges(G, pos, edgelist=gene_TF_edges_neg, edge_color='silver',  width=0.5, alpha=0.5, style="dashed")


#curved_graph(G,pos)

ax = plt.gca()

for i, row in TF_cols.iterrows():
    ax.scatter(None,None, label=row['TFFs'], color=row['color'])

ax.legend(loc=5)

ax.margins(0.15)
plt.axis("off")
plt.tight_layout()


#plt.savefig("multi_circle_network_style2.pdf")

import bezier
from matplotlib.collections import LineCollection
import random

def curved_graph(graph, pos = None, eps=0.2, pointn=30):

    if pos == None:
        pos = nx.spring_layout(graph)

    for u,v in graph.edges():
        x0, y0 = pos[u]
        x1, y1 = pos[v]
        segs = curved_line(x0,y0,x1,y1)
        lc = LineCollection(segs)
        plt.gca().add_collection(lc)
        plt.gca().autoscale_view()

def curved_line(x0, y0, x1, y1, eps=0.2, pointn=30):

    x2 = (x0+x1)/2.0 + 0.1 ** (eps+abs(x0-x1)) * (-1) ** (random.randint(1,4))
    y2 = (y0+y1)/2.0 + 0.1 ** (eps+abs(y0-y1)) * (-1) ** (random.randint(1,4))
    nodes = np.asfortranarray([
        [x0, x2, x1],
        [y0, y2, y1]
    ])
    curve = bezier.Curve(nodes,
                         degree=2)
    s_vals = np.linspace(0.0, 1.0, pointn)
    data=curve.evaluate_multi(s_vals)
    x=data[0]
    y=data[1]
    segments =[]
    for index in range(0,len(x)):
        segments.append([x[index],y[index]])
    segments = [segments]
    return  segments

def get_nodes(df, scale=1, center=np.array([0,0])):
    G = nx.cycle_graph(df.shape[0])
    pos = nx.circular_layout(G, scale=scale, center=center)
    pos = {df['gene'].to_list()[i]:pos[i] for i in pos}

    return G, pos

plt.figure(1, figsize=(10, 9))



c23 = pd.concat([c2, c3], axis=0)


c23['TFfamily'] = None

for i,row in TF_cols.iterrows():
    c23.loc[c23[c23['gene'].str.contains(row['TFFs'])].index, "TFfamily"] = row['TFFs']

c23['TFcolor'] = TF_cols.loc[c23['TFfamily'], 'color'].to_list()
c23 = c23.sort_values(['corr', 'TFfamily'])



G1, pos1 = get_nodes(c1, scale=2, center=np.array([0,0]))
G23, pos23 = get_nodes(c23, scale=10, center=np.array([0,0]))




G = nx.Graph()
G.clear()

for i in c1['gene'].to_list():
    G.add_node(i)

for i in c23['gene'].to_list():
    G.add_node(i, )

G.add_node("Suc")


pos1_x = [5,4,3,2,1,-1,-2,-3,-4,-5]
pos1_y = [3,4,5,6,7,7,6,5,4,3]
pos1 = {v:np.array([pos1_x[i], pos1_y[i]]) for i,v in enumerate(c1['gene'].to_list())}

pos = dict(pos1, **pos23)
pos['Suc'] = np.array([0,-4])


nx.draw_networkx_nodes(G, {'Suc': pos['Suc']}, nodelist=['Suc'], node_size=500, node_color='#C7442A')
nx.draw_networkx_nodes(G, pos1, nodelist=c1['gene'].to_list(), node_size=300, node_color='#E0CC71')


for i in c23['TFfamily'].unique():
    nx.draw_networkx_nodes(G, pos23, nodelist=c23[c23['TFfamily']==i]['gene'].to_list(), node_size=200, node_color=TF_cols.loc[i, 'color'])


nx.draw_networkx_labels(G, {'Suc': pos['Suc']}, labels={'Suc': 'Suc'}, font_size=12, font_family="sans-serif", font_color='w')


nx.draw_networkx_labels(G, pos1, labels={i:i for i in pos1}, font_size=10, font_family="sans-serif", font_color='k')
#nx.draw_networkx_labels(G, pos23, labels={i:i for i in pos23}, font_size=6, font_family="sans-serif", font_color='k')




def draw_edge(df, G, pos, pos_edge_color='#E00671', neg_edge_color='#E00671', width=5, alpha=1):
    # add edges
    #c_edges = []
    c_edges_pos = []
    c_edges_neg = []
    for i,row in df.iterrows():
        G.add_edge(row['gene'], row['Sugar'], length=row['absR'], weight=row['corr'])
        #c_edges.append((row['gene'], row['Sugar']))
        if row['corr'] > 0:
            c_edges_pos.append((row['gene'], row['Sugar']))
        else:
            c_edges_neg.append((row['gene'], row['Sugar']))


    nx.draw_networkx_edges(G, pos, edgelist=c_edges_pos, edge_color=pos_edge_color, width=width*1.2, alpha=1, style = 'solid')
    nx.draw_networkx_edges(G, pos, edgelist=c_edges_neg, edge_color=neg_edge_color, width=width, alpha=0.6, style = 'dashed')

draw_edge(c1, G, pos, pos_edge_color='#E00671', neg_edge_color='#E00671', width=2)
draw_edge(c23, G, pos, pos_edge_color='#E0CCD2', neg_edge_color='g', width=1, alpha=0.5)


#gene_TF_edges = []
gene_TF_edges_pos = []
gene_TF_edges_neg = []

for i,row in gene_TF_corr.query('absR>0.6').iterrows():
    G.add_edge(row['Gene_symbol'], row['TF_symbol'], length=row['absR'], weight=row['corr'])
    #gene_TF_edges.append((row['Gene_symbol'], row['TF_symbol']))
    if row['corr'] > 0:
        gene_TF_edges_pos.append((row['Gene_symbol'], row['TF_symbol']))
    else:
        gene_TF_edges_neg.append((row['Gene_symbol'], row['TF_symbol']))


nx.draw_networkx_edges(G, pos, edgelist=gene_TF_edges_pos, edge_color='#A0CBE2',  width=1, alpha=0.5 )
nx.draw_networkx_edges(G, pos, edgelist=gene_TF_edges_neg, edge_color='silver',  width=1, alpha=1, style="dashed")


#curved_graph(G,pos)

ax = plt.gca()

for i, row in TF_cols.iterrows():
    ax.scatter(None,None, label=row['TFFs'], color=row['color'])

ax.legend(loc=5)

ax.margins(0.15)
plt.axis("off")
plt.tight_layout()


#plt.savefig("multi_circle_network_style3.pdf")

gene_TF_corr

	Unnamed: 0	Gene	TF	corr	absCorr	Gene_symbol	TF_symbol
0	979	Solyc09g098590.4	Solyc04g077490.3	0.223526	0.223526	SlSUSY2	SlERF3
1	980	Solyc09g098590.4	Solyc01g067560.3	-0.389336	0.389336	SlSUSY2	SlMYB14
2	981	Solyc09g098590.4	Solyc10g080770.3	-0.487784	0.487784	SlSUSY2	SlZIP4
3	982	Solyc09g098590.4	Solyc02g063430.4	0.184914	0.184914	SlSUSY2	SlHLH2
4	983	Solyc09g098590.4	Solyc02g076810.4	-0.502613	0.502613	SlSUSY2	SlTrihelix5
...	...	...	...	...	...	...	...
785	7648	Solyc08g042000.3	Solyc02g087920.3	-0.645652	0.645652	SlSPS1	SlNAC7
786	7649	Solyc08g042000.3	Solyc07g047610.3	0.404265	0.404265	SlSPS1	SlMYB2
787	7651	Solyc08g042000.3	Solyc09g057710.4	0.521755	0.521755	SlSPS1	SlHLH1
788	7652	Solyc08g042000.3	Solyc03g078120.4	-0.589939	0.589939	SlSPS1	SlNAC6
789	7653	Solyc08g042000.3	Solyc08g078340.3	-0.574457	0.574457	SlSPS1	SlMYB18

790 rows × 7 columns

def plot_df_cir(df, scale=1, cmap=plt.cm.Blues):
    G = nx.cycle_graph(df.shape[0])
    pos = nx.circular_layout(G, scale=scale)

    #nx.draw(G, pos, node_color=, node_size=800, cmap=cmap)
    labels = {n:df['gene'].to_list()[n] for n,lab in G.nodes(data=True)}
    nodes = nx.draw_networkx_nodes(G, pos, node_size=500, node_color=df['absR'], cmap=cmap, alpha=0.5)
    nx.draw_networkx_labels(G, pos, font_size=8,
        font_family="sans-serif", labels=labels, verticalalignment='center')

    #pc = mpl.collections.PatchCollection(nodes, cmap=cmap)
    #plt.colorbar(pc)

    return nodes

plt.figure(1, figsize=(9, 9))

G1 = plot_df_cir(c1, scale=3, cmap=plt.cm.Reds)
G2 = plot_df_cir(c2, scale=8, cmap=plt.cm.Blues)
G3 = plot_df_cir(c3, scale=10, cmap=plt.cm.Purples)

#G = nx.Graph()
#nx.draw_networkx_nodes(G, {0: np.array([0,0])}, node_size=500, node_color=['red'])



ax = plt.gca()
ax.margins(0.15)
plt.axis("off")
plt.tight_layout()


plt.savefig("multi_circle_network.pdf")