Circos plotting the CCA results

I run CCA on a set of variables and got 4 components significant after fdr correction. Each components got top vars extracted by thresholding the cca weight at 0.2 and now I wanted to plot the components and top vars in each fc and sc as seen bellow of the code:

import os
import networkx as nx
import numpy as np
import pandas as pd
from sklearn.cross_decomposition import CCA
from scipy.stats import percentileofscore
from statsmodels.stats.multitest import multipletests
from sklearn.utils import resample
from nxviz import CircosPlot
import matplotlib.pyplot as plt
from sklearn.preprocessing import StandardScaler


#Set seed
np.random.seed(123)

# Load data
def load_data(file_name):
    return pd.read_csv(file_name)

# Canonical Correlation Analysis
def calculate_canonical_loadings(X, Y):
    cca = CCA(n_components=69, max_iter=5000)
    cca.fit(X, Y)
    X_c, Y_c = cca.transform(X, Y)
    return np.corrcoef(X_c.T, Y_c.T)[:X_c.shape[1], X_c.shape[1]:], cca

# Bootstrap for CCA coefficients
def bootstrap_cca_coefficients(X, Y, n_iterations=1000, alpha=0.05):
    bootstrap_coefs = []
    cca = CCA(n_components=69, max_iter=5000)
    for _ in range(n_iterations):
        resampled_X, resampled_Y = resample(X, Y)
        cca.fit(resampled_X, resampled_Y)
        bootstrap_coefs.append(cca.x_weights_)
    lower, upper = np.percentile(bootstrap_coefs, [alpha/2*100, (1-alpha/2)*100], axis=0)
    return lower, upper

def permutation_test_cca(X, Y, observed_corr, n_iterations=1000):
    p_values = np.zeros(observed_corr.shape[0])
    permuted_corr = np.zeros((n_iterations, observed_corr.shape[0]))
    cca = CCA(n_components=69, max_iter=5000)
    for i in range(n_iterations):
        shuffled_X = np.random.permutation(X)
        shuffled_Y = np.random.permutation(Y)
        cca.fit(shuffled_X, shuffled_Y)
        X_c, Y_c = cca.transform(shuffled_X, shuffled_Y)
        permuted_corr[i] = np.diag(np.corrcoef(X_c.T, Y_c.T)[:X_c.shape[1], X_c.shape[1]:])
    p_values = 1 - np.array([percentileofscore(permuted_corr[:, i], observed_corr[i]) / 100 for i in range(observed_corr.shape[0])])
    return p_values
 
# Extract top variable names
def get_top_variable_names_for_significant_components(data, weights, significant_indices, threshold=0.2):
    top_variable_names = {}
    column_names = data.columns
    for index in significant_indices:
        weight_vector = weights[:, index]
        top_variable_indices = np.where(np.abs(weight_vector) >= threshold)[0]
        top_variable_names[index] = column_names[top_variable_indices].tolist()
    return top_variable_names

# Function to prepare CCA data for Circos plot
def prepare_cca_data_for_circos(top_fc_variable_names, top_sc_variable_names, observed_corr, significant_fdr):
    edge_list = []
    for comp in significant_fdr:
        for fc_var in top_fc_variable_names[comp]:
            for sc_var in top_sc_variable_names[comp]:
                edge_list.append((fc_var, sc_var, observed_corr[comp, comp]))
    return edge_list

# Function to plot Circos plot
def plot_circos(edge_list):
    G = nx.Graph()
    for fc_var, sc_var, weight in edge_list:
        G.add_edge(fc_var, sc_var, weight=weight)
    
    c = CircosPlot(G, node_color="class", node_order="class", node_labels=True)
    c.draw()
    plt.show()

# Main function
if __name__ == "__main__":
    desktop_path = os.path.expanduser("~/Desktop/struc_func/The_NMF_Run_data")
    os.chdir(desktop_path)
    
    fc_data = load_data("top_vars_nmf_struc.csv")
    sc_data = load_data("top_vars_nmf_func.csv")
    
    scaler = StandardScaler()
    fc_data_standardized = scaler.fit_transform(fc_data)
    sc_data_standardized = scaler.fit_transform(sc_data)
    
    observed_corr, cca = calculate_canonical_loadings(fc_data_standardized, sc_data_standardized)
    p_values = permutation_test_cca(fc_data_standardized, sc_data_standardized, np.diag(observed_corr))
        
    reject, pvals_corrected, _, _ = multipletests(p_values, alpha=0.05, method='fdr_bh')
    significant_fdr = np.where(pvals_corrected < 0.05)[0]
    
    if len(significant_fdr) > 0:
        top_fc_variable_names = get_top_variable_names_for_significant_components(fc_data, cca.x_weights_, significant_fdr, threshold=0.2)
        top_sc_variable_names = get_top_variable_names_for_significant_components(sc_data, cca.y_weights_, significant_fdr, threshold=0.2)

        edge_list = prepare_cca_data_for_circos(top_fc_variable_names, top_sc_variable_names, observed_corr, significant_fdr)
        
        plot_circos(edge_list)
    else:
        print("No significant components found after FDR correction.")
  
Plotting (circos)
 def generate_circos_plot_matplotlib(nodes, links, top_fc_variable_names=None, top_sc_variable_names=None):
    fig, ax = plt.subplots(figsize=(10, 10), subplot_kw={'projection': 'polar'})
    N = len(nodes)
    
    # Sort the nodes by their 'color' property for better visualization
    sorted_nodes = sorted(nodes, key=lambda x: x['color'])
    
    # Generate theta values
    theta = np.linspace(0.0, 2 * np.pi, N, endpoint=False)
    
    # Create the node points on the circle
    for i, node in enumerate(sorted_nodes):
        ax.scatter(theta[i], 1, c=node['color'], s=300)
        ax.text(theta[i], 1.1, node['label'], ha='center')
        
    # Add logic to highlight top variables (if provided)
    if top_fc_variable_names and top_sc_variable_names:
        # Your logic to highlight top variables goes here.
        pass
        
    # Create links
    for link in links:
        source_idx = next((i for i, node in enumerate(sorted_nodes) if node['id'] == link['source']), None)
        target_idx = next((i for i, node in enumerate(sorted_nodes) if node['id'] == link['target']), None)
        
        if source_idx is None or target_idx is None:
            continue
        
        # Convert the weights to color
        color_intensity = np.clip(link['value'], 0, 1)
        color = (1 - color_intensity, 0.2, color_intensity)
        
        ax.plot([theta[source_idx], theta[target_idx]], [1, 1], c=color, lw=2*link['value'])
        
    plt.show()

But I was not successful in achieving my goal, I know the I ran the CCA correctly. I appreciate any help in plotting this.

source https://stackoverflow.com/questions/77137376/circos-plotting-the-cca-results