Have you ever considered how a family tree might map onto a phylogenetic diagram, with each branch representing both social links and genetic information? I’d love to compare the data structures we use.

Sounds like a fascinating project—are you looking at a specific ethnic group, or a broader dataset of multiple clans? I could help with aligning the kinship matrices to the genetic distance tables and checking for correlations.

Let’s start by normalizing each kinship matrix so that entries are the number of shared ancestors, then compute the pairwise genetic distance from the mitochondrial haplogroup frequencies. We can stack the two matrices into a single table and run a Mantel test to see if the correlation holds. I can script the alignment in R or Python—just let me know which you prefer. Once we have the paired distances, we’ll map the taboo links onto the graph and see if they sit at the outliers. Ready to code?

Sounds solid—just let me know when you’ve got the matrices ready, and I’ll start the script to compute the pairwise genetic distances, run the Mantel test, and build the network graph with the taboo edges highlighted. We can iterate on any assumptions as we go. Ready when you are.

Here’s a quick script that pulls everything together. Copy it into a .py file and run it after you’ve put the two CSVs in the same folder. It assumes the kinship matrix has rows/cols labelled by clan, and the haplogroup table has one row per clan with columns for each haplogroup frequency (they should sum to 1). ```python import pandas as pd import numpy as np from scipy.spatial.distance import pdist, squareform from skbio.stats.distance import mantel import networkx as nx import matplotlib.pyplot as plt # Load data kinship = pd.read_csv('kinship_matrix.csv', index_col=0) # [clan, clan] haplo = pd.read_csv('haplogroup_frequencies.csv', index_col=0) # [clan, haplogroup] # Ensure the same ordering clans = kinship.index.intersection(haplo.index) kinship = kinship.loc[clans, clans] haplo = haplo.loc[clans] # -------------------- 1. Genetic distances -------------------- # Euclidean distance between haplogroup frequency vectors gen_dist = pd.DataFrame( squareform(pdist(haplo, metric='euclidean')), index=clans, columns=clans ) # -------------------- 2. Mantel test -------------------- # Convert to condensed distance matrices for the Mantel function kinship_cond = pdist(kinship, metric='euclidean') gen_cond = pdist(gen_dist, metric='euclidean') mantel_stat, p_value = mantel(kinship_cond, gen_cond, method='pearson', permutations=9999) print(f'Mantel statistic: {mantel_stat:.4f}, p‑value: {p_value:.4f}') # -------------------- 3. Build sociogram graph -------------------- G = nx.Graph() G.add_nodes_from(clans) # Add all edges weighted by kinship for i, clan_i in enumerate(clans): for j, clan_j in enumerate(clans): if i < j: G.add_edge(clan_i, clan_j, weight=kinship.loc[clan_i, clan_j]) # Mark taboo edges (replace your list of taboo pairs) taboo_edges = [('Trobriand', 'Bwanabwana'), ('Melanesian1', 'Melanesian2')] for u, v in taboo_edges: if G.has_edge(u, v): G[u][v]['taboo'] = True else: G.add_edge(u, v, taboo=True) # -------------------- 4. Plot -------------------- pos = nx.spring_layout(G, seed=42) edge_colors = ['red' if G[u][v].get('taboo') else 'grey' for u, v in G.edges()] edge_widths = [G[u][v]['weight'] for u, v in G.edges()] nx.draw_networkx_nodes(G, pos, node_color='lightblue', node_size=700) nx.draw_networkx_labels(G, pos, font_size=12, font_family='sans-serif') nx.draw_networkx_edges(G, pos, edge_color=edge_colors, width=edge_widths, alpha=0.7) plt.title('Kinship sociogram – taboo edges in red') plt.axis('off') plt.tight_layout() plt.show() ``` A couple of points to double‑check: 1. If your kinship counts are “number of shared grandparents” or something, you might want to transform them (e.g., 1 / (count + 1)) so that larger numbers mean closer kinship, which aligns with the intuition that small genetic distances should match high kinship. 2. The Mantel test uses Euclidean distances on the raw matrices; if you think a different metric (e.g., log‑ratio or Jaccard) is more appropriate, swap the `metric` argument. Run the script, inspect the Mantel statistic and p‑value, and look at the plotted graph. If the red taboo edges cluster in the high‑weight, low‑genetic‑distance region, that’s a neat hint that cultural rules were probably reacting to genetic closeness. If they’re scattered, the story might be more sociological than genetic. Happy coding!