Skip to content

SKILL.md

Latest commit

 

History

History
224 lines (167 loc) · 5.89 KB

SKILL.md

File metadata and controls

224 lines (167 loc) · 5.89 KB
name bio-tcr-bcr-analysis-repertoire-visualization
description Create publication-quality visualizations of immune repertoire data including circos plots, clone tracking, diversity plots, and network graphs. Use when generating figures for repertoire comparisons, clonal dynamics, or V(D)J gene usage.
tool_type mixed
primary_tool VDJtools

Version Compatibility

Reference examples tested with: MiXCR 4.6+, VDJtools 1.2.1+, ggplot2 3.5+, matplotlib 3.8+, pandas 2.2+, scanpy 1.10+, seaborn 0.13+

Before using code patterns, verify installed versions match. If versions differ:

  • Python: pip show <package> then help(module.function) to check signatures
  • R: packageVersion('<pkg>') then ?function_name to verify parameters
  • CLI: <tool> --version then <tool> --help to confirm flags

If code throws ImportError, AttributeError, or TypeError, introspect the installed package and adapt the example to match the actual API rather than retrying.

Repertoire Visualization

"Visualize my immune repertoire data" → Create publication-quality figures for TCR/BCR repertoires including circos plots, V(D)J gene usage heatmaps, diversity plots, and clonal tracking across samples.

  • CLI: vdjtools PlotFancyVJUsage for circos-style V-J plots
  • Python: matplotlib/seaborn for custom repertoire visualizations

Circos Plots (V-J Gene Usage)

VDJtools

# Generate V-J usage circos plot
vdjtools PlotFancyVJUsage \
    -m metadata.txt \
    output_dir/

# Generates PDF circos plots showing V-J pairing frequencies

Python with pyCircos

import pandas as pd
import matplotlib.pyplot as plt
from pycircos import Gcircle

def plot_vj_circos(clone_df):
    '''Create circos plot of V-J usage'''
    # Count V-J pairs
    vj_counts = clone_df.groupby(['v_gene', 'j_gene']).size().reset_index(name='count')

    # Create circos
    circle = Gcircle()

    # Add arcs for each V and J gene
    v_genes = vj_counts['v_gene'].unique()
    j_genes = vj_counts['j_gene'].unique()

    # Add sectors and links
    # ... (complex setup)

    circle.save('vj_circos.pdf')

R with circlize

library(circlize)

plot_vj_circos <- function(clone_df) {
    # Prepare adjacency matrix
    vj_matrix <- table(clone_df$v_gene, clone_df$j_gene)

    # Create circos plot
    chordDiagram(
        vj_matrix,
        transparency = 0.5,
        annotationTrack = c("grid", "name")
    )
}

Clone Tracking Over Time

import pandas as pd
import matplotlib.pyplot as plt

def plot_clone_tracking(clones_by_time, top_n=10):
    '''Track top clones across timepoints'''

    # Get top clones by total frequency
    total_freq = clones_by_time.groupby('cdr3_aa')['frequency'].sum()
    top_clones = total_freq.nlargest(top_n).index

    fig, ax = plt.subplots(figsize=(10, 6))

    for clone in top_clones:
        clone_data = clones_by_time[clones_by_time['cdr3_aa'] == clone]
        ax.plot(clone_data['timepoint'], clone_data['frequency'],
                marker='o', label=clone[:20])

    ax.set_xlabel('Timepoint')
    ax.set_ylabel('Clone Frequency')
    ax.legend(bbox_to_anchor=(1.05, 1), loc='upper left')
    plt.tight_layout()
    plt.savefig('clone_tracking.pdf')

Diversity Plots

import matplotlib.pyplot as plt
import seaborn as sns

def plot_diversity_comparison(diversity_df, metric='shannon'):
    '''Compare diversity between groups'''

    fig, ax = plt.subplots(figsize=(8, 6))

    sns.boxplot(
        data=diversity_df,
        x='condition',
        y=metric,
        ax=ax
    )
    sns.stripplot(
        data=diversity_df,
        x='condition',
        y=metric,
        color='black',
        alpha=0.5,
        ax=ax
    )

    ax.set_ylabel(f'{metric.capitalize()} Diversity')
    plt.savefig('diversity_comparison.pdf')

Overlap Heatmap

def plot_overlap_heatmap(overlap_matrix):
    '''Plot pairwise repertoire overlap'''
    import seaborn as sns

    fig, ax = plt.subplots(figsize=(10, 8))

    sns.heatmap(
        overlap_matrix,
        annot=True,
        fmt='.2f',
        cmap='YlOrRd',
        ax=ax
    )

    ax.set_title('Repertoire Overlap (Jaccard Index)')
    plt.tight_layout()
    plt.savefig('overlap_heatmap.pdf')

Spectratype Plot

def plot_spectratype(clone_df, group_col=None):
    '''Plot CDR3 length distribution'''

    fig, ax = plt.subplots(figsize=(10, 6))

    clone_df['cdr3_length'] = clone_df['cdr3_nt'].str.len()

    if group_col:
        for group, data in clone_df.groupby(group_col):
            ax.hist(data['cdr3_length'], bins=range(20, 80, 3),
                    alpha=0.5, label=group, density=True)
        ax.legend()
    else:
        ax.hist(clone_df['cdr3_length'], bins=range(20, 80, 3))

    ax.set_xlabel('CDR3 Length (nt)')
    ax.set_ylabel('Density')
    ax.set_title('CDR3 Length Distribution (Spectratype)')
    plt.savefig('spectratype.pdf')

Clonotype Network

import networkx as nx

def plot_clone_network(clone_df, similarity_threshold=0.8):
    '''Create network of similar clonotypes'''
    from Levenshtein import ratio

    G = nx.Graph()

    clones = clone_df['cdr3_aa'].unique()

    # Add nodes
    for clone in clones:
        freq = clone_df[clone_df['cdr3_aa'] == clone]['frequency'].sum()
        G.add_node(clone, size=freq)

    # Add edges for similar clones
    for i, c1 in enumerate(clones):
        for c2 in clones[i+1:]:
            sim = ratio(c1, c2)
            if sim >= similarity_threshold:
                G.add_edge(c1, c2, weight=sim)

    # Draw network
    fig, ax = plt.subplots(figsize=(12, 12))
    pos = nx.spring_layout(G)

    sizes = [G.nodes[n]['size'] * 1000 for n in G.nodes()]
    nx.draw(G, pos, node_size=sizes, with_labels=False, ax=ax)

    plt.savefig('clone_network.pdf')

Related Skills

  • vdjtools-analysis - Generate input data
  • mixcr-analysis - Generate clonotype tables
  • data-visualization/ggplot2-fundamentals - General plotting concepts