AE-Trans Integrates Unpaired Multi-Omics For Diagnosis
Researchers (Liao et al.) published March 12, 2026 in PLoS Computational Biology introduce AE-Trans, an interpretable dual-channel Transformer that integrates unpaired transcriptomic and DNA methylation data using cross-modal reconstruction to diagnose Alzheimer’s disease. AE-Trans achieved 0.9736 accuracy and 0.9910 AUC on prefrontal cortex cohorts, generalized to external cohorts (accuracy 0.7389, AUC 0.8432), and identified pathway-linked biomarkers validated by enrichment and logistic regression.
Key Points
- 1Achieves high diagnostic performance: 0.9736 accuracy and 0.9910 AUC on prefrontal cortex cohorts.
- 2Aligns unpaired RNA and DNA methylation via cross-modal transformer reconstruction for interpretable inter-omic relationships.
- 3Provides validated biomarkers and latent embeddings that stratify patients and improve prognostic prediction in AD.
Scoring Rationale
Peer-reviewed transformer-based novelty and strong validation drive score, limited by domain focus and reduced external cohort accuracy.
Sources
Public references used for this report.
Practice interview problems based on real data
1,625 SQL & Python problems across 15 industry datasets — the exact type of data you work with.
Try 250 free problems
