kmlyyll/epibarrett-simulated-cohort

epibarrett simulated cohort

A biologically calibrated, synthetic HumanMethylation450-style dataset for Barrett's esophagus / esophageal adenocarcinoma detection.

Files

• X.parquet — samples × probes beta-value matrix (460 samples, 2074 probes)
• meta.parquet — sample metadata: label, stage, cohort, age, sex_male, bmi, smoker, gerd

Cohorts

• GSE81334_like — balanced discovery cohort (n=250, 50% cases)
• GSE104707_like — screening-like external cohort (n=192, ~8% cases)

Usage

import pandas as pd

X = pd.read_parquet("X.parquet")
meta = pd.read_parquet("meta.parquet")

Note

This is synthetic data for pipeline demonstration and CI. It is not a validated clinical dataset. For real GEO cohorts see the epibarrett GitHub repository.