DNA methylation visualizations using GEO datasets

This example demonstrates the built-in DNA methylation plotting function to visualize DNA methylation against age across datasets in a violin and linear and polynomial regression plots.

Import required classes, and functions

from biolearn.data_library import DataLibrary
from biolearn.visualize import (
    identify_stable_cpg_sites,
    compute_age_group_stats,
    compute_methylation_stats,
    plot_methylation_by_age_sex,
    plot_methylation_vs_age
)

Load GEO methylation dataset(s) and set up objects for visualization function use

# List dataset IDs
dataset_ids = ["GSE112618", "GSE110554", "GSE41169", "GSE52588"]

# Load datasets into a dictionary of GeoData objects
library = DataLibrary()
datasets = {id: library.get(id).load() for id in dataset_ids}

Identify the top 10 stable CpG sites with low variance for each dataset

stable_sites = identify_stable_cpg_sites(datasets, threshold=0.01)
for dataset in stable_sites:
    dataset_name = dataset["Dataset"]
    stable_cpg_sites = dataset["Stable CpG Sites"][:10]
    print(f"{dataset_name}: {stable_cpg_sites}")

GSE112618: ['cg00000029', 'cg00000103', 'cg00000109', 'cg00000155', 'cg00000158', 'cg00000165', 'cg00000221', 'cg00000236', 'cg00000289', 'cg00000292']
GSE110554: ['cg00000103', 'cg00000109', 'cg00000155', 'cg00000158', 'cg00000165', 'cg00000221', 'cg00000289', 'cg00000292', 'cg00000321', 'cg00000363']
GSE41169: ['cg00000029', 'cg00000108', 'cg00000109', 'cg00000165', 'cg00000236', 'cg00000289', 'cg00000292', 'cg00000321', 'cg00000363', 'cg00000622']
GSE52588: ['cg00000029', 'cg00000108', 'cg00000109', 'cg00000165', 'cg00000236', 'cg00000289', 'cg00000292', 'cg00000321', 'cg00000363', 'cg00000622']

Visualize DNA methylation levels by age and sex in a violin plot

# Initialize an appropriate CpG site
cpg_site = "cg00000029"

# Compute statistics by age group for the specified CpG site
stats_df, combined_data1 = compute_age_group_stats(datasets, cpg_site, age_bins=5)

# Plot DNA methylation levels by age group and sex across datasets in a violin plot
# Note: In the datasets, '1' represents Female and '2' represents Male.
plot_methylation_by_age_sex(combined_data1, cpg_site)

Visualize DNA methylation against age with linear and polynomial regression

# Compute regression statistics to plot methylation vs. age
combined_data2, stats_df = compute_methylation_stats(datasets, cpg_site)

# Plot DNA methylation against age across datasets with linear and polynomial regression lines
plot_methylation_vs_age(combined_data2, stats_df, cpg_site)

Computed statistics for CpG site: cg00000029
Linear Regression R^2: 0.0311
Pearson Correlation (r): -0.1764
Pearson Correlation p-value: 8.2813e-03
Polynomial Regression (Degree 2) R^2: 0.0327

Estimated memory usage: 1814 MB