DNA methylation

Schizophrenia DNA Methylation (GSE147221)

Blood DNA methylation profiles from the Dublin schizophrenia study, measured at CpG sites with Illumina HumanMethylation450 arrays. The task is to distinguish participants with schizophrenia from control participants using their methylation measurements.

Binary classificationLicence: NCBI GEO public data; study-specific terms require verification before re-hostingGEO-GSE147221-Methylation-SchizophreniaTarget: statusgeo_matrix ↗

At this operating point

Model comparison

p=10,000, n=100 · Higher Macro-F1 is better

Random Forest is marked with a diamond. Models are ordered by the selected metric.

Across sample budgets

Learning curves

Models

All metric values at this operating point

Model	Macro-F1	MCC	Balanced accuracy	ROC-AUC
RealTabPFN v2	0.8626	0.7412	0.8626	0.9361
LimiX2	0.8582	0.7191	0.8587	0.9358
Causilo	0.8522	0.7092	0.8524	0.9378
TabPFN Wide 5k (ne3)	0.8509	0.7044	0.8508	0.9335
Random Forest (tuned)	0.8434	0.6918	0.8439	0.9189
MITRA	0.8334	0.6676	0.8337	0.9067
Kumo Tabular (M)	0.8329	0.67	0.8328	0.9059
Extra Trees (tuned)	0.8228	0.6509	0.8237	0.9108
TabPFN 3.5	0.8224	0.6504	0.8226	0.91
RealTabPFN 2.5	0.822	0.651	0.8222	0.9031
TabFM	0.8152	0.6558	0.818	0.9151
XGBoost (tuned)	0.8038	0.6171	0.8045	0.8449
XGBoost	0.7898	0.5868	0.7907	0.8557
GeneICL	0.7746	0.5565	0.7755	0.8351
Logistic Regression (tuned)	0.7725	0.5539	0.7735	0.8344
Logistic Regression	0.7701	0.5468	0.7706	0.8385
AutoGluon	0.7513	0.5211	0.7536	0.8289
Random Forest	0.748	0.5024	0.749	0.8274
Extra Trees	0.745	0.4965	0.7461	0.8062
TabM	0.7397	0.4848	0.7405	0.8194
RealMLP	0.7332	0.4834	0.7358	0.8307
MLP	0.7118	0.4325	0.7138	0.7881
MLP (tuned)	0.7019	0.4216	0.7062	0.7871
TabDPT	0.6839	0.4645	0.7025	0.8297