Hypothesis

Adding a fourth epigenomic layer—specifically, DNA methylation quantitative trait loci (meQTLs)—to existing multi‑omics pipelines converts correlative associations into directional causal inferences, thereby improving disease subtyping, biomarker discovery, and prediction of therapeutic response beyond what genomics‑proteomics‑metabolomics alone can achieve.

Rationale

Current multi‑omics approaches excel at capturing cross‑layer correlations but struggle to infer causality [7]. Epigenetic modifications sit mechanistically between genetic variation and downstream protein/metabolite abundances; meQTLs link genotypes to methylation states, which in turn influence gene expression and protein levels. By explicitly modeling these genotype‑>methylation>proteom/metabolite chains, we can anchor observed multi‑omics changes to upstream genetic drivers, satisfying Mendelian randomization criteria for causality.

MeQTL data are increasingly accessible through large consortia (e.g., GOGT, BIOS QTLs) and can be integrated with existing proteogenomic frameworks that already combine proteomics and Mendelian randomization [6]. Deep‑learning multi‑view autoencoders can accommodate the additional heterogeneity [3], while similarity network fusion preserves sample relationships even with missing methylation values [4].

Testable Predictions

1. Improved risk prediction – In an independent cohort of Alzheimer’s disease patients, a model trained on genomics + proteomics + metabolomics + meQTLs will achieve an AUROC ≥ 0.78, a statistically significant increase over the 0.703 AUROC reported for transcriptomics‑plus‑covariates random forest [1].
2. Causal biomarker identification – Proteins whose abundance is significantly mediated by methylation (i.e., significant genotype→methylation→protein path) will show enrichment for known drug targets and will be validated by CRISPR‑dCas9 methylation editing in vitro, resulting in measurable changes in protein level and downstream metabolite flux.
3. Treatment response stratification – In glioma, patients stratified by multi‑omics risk scores that include meQTL‑derived causal proteins will exhibit a ≥ 20 % difference in progression‑free survival compared to stratification using genomics‑proteomics‑metabolomics alone [2].

Experimental Design

• Data collection – Obtain matched germline genotype, whole‑blood or tissue DNA methylation, bulk proteomics (LC‑MS/MS), and untargeted metabolomics from 500 Alzheimer’s cases and 500 controls (or 300 glioma patients with treatment outcomes).
• Processing – Call meQTLs using MatrixEQTL, normalize each layer with combat‑seq, and impute missing methylation values via k‑nearest neighbours.
• Modeling – Train a multi‑view autoencoder to learn joint latent representations; attach a downstream gradient‑boosted classifier for outcome prediction. Compare AUROC against baseline models lacking the methylation view using DeLong’s test.
• Mediation analysis – Apply causal mediation analysis (R package ‘mediation’) to test genotype→methylation→protein paths; retain proteins with significant indirect effects (p < 0.01 after FDR correction).
• Validation – Select top three causal proteins; use CRISPR‑dCas9‑TET1 or dCas9‑DNMT3A to demethylate/methylate promoter CpGs in iPSC‑derived neurons or glioma lines; measure protein abundance (Western blot) and metabolite shifts (targeted LC‑MS).

Potential Pitfalls & Mitigations

• Confounding by cell‑type composition – Adjust methylation data with reference‑based deconvolution (e.g., MeDeCom) before meQTL mapping.
• Horizontal pleiotropy – Use MR‑Egger and weighted median methods to assess pleiotropic bias in mediation estimates.
• Model overfitting – Employ nested cross‑validation and hold‑out test set; limit latent dimensions via explained variance threshold.

If the predictions hold, this work demonstrates that epigenomic layering converts multi‑omics from a pattern‑finding exercise into a causal discovery engine, directly addressing the field’s foremost bottleneck [7] and moving biomarker identification toward actionable therapeutic targets.