Experiments

Searchable full-text extractions: founding hypothesis, core claims, experimental setups, key results and statistics — pulled out of each paper as structure. Search a cell line, an assay or an entity (e.g. HUH7) and find every paper that worked with it. This corpus stands on its own: most entries carry no reproduction assessment (yet).

• Lorentz-regularized interpretable VAE for multi-scale single-cell transcriptomic and epigenomic embeddings.

  PMID 41555918 · PMC12812404 · Frontiers in genetics · 2025 · 7 claims · 5 setups

  Full-text index only

  LiVAE, a dual-pathway VAE with Lorentzian geometric regularization between a primary Euclidean pathway and an information-bottleneck pathway, balances local fidelity with global topology coherence in single-cell embeddings

  single-cell same dapp1 benchmarking umap component-wise

• Machine learning-predicted chromatin organization landscape across pediatric tumors.

  PMID 41904260 · PMC13039956 · Scientific reports · 2026 · 8 claims · 5 setups

  Full-text index only

  SuPreMo-Akita (built on the Akita CNN) enables systematic in silico prediction of somatic SV effects on 3D genome folding across large SV cohorts where experimental testing is infeasible

  somatic per-sample tumor paired mutagenesis dimensionality h3k27ac

• A unified framework for correcting batch effects and integrating multi-omics data.

  PMID 41786846 · PMC13079841 · Scientific reports · 2026 · 7 claims · 6 setups

  Full-text index only

  MoDAmix, a four-stage domain adaptation framework (pre-training, single-omics adversarial adaptation, multi-omics adversarial alignment, semi-supervised class alignment), unifies batch correction across multiple omics layers.

  brain aml same single-cell bulk k-means

• iAODE for benchmarking and continuum modeling of single-cell chromatin accessibility.

  PMID 41775921 · PMC13066597 · Communications biology · 2026 · 8 claims · 5 setups

  Full-text index only

  iAODE combines a ZINB-likelihood VAE, a latent Neural ODE, low-weight KL regularization, and an interpretable reconstruction (irecon) bottleneck to learn generative, temporally continuous latent spaces for scATAC-seq.

  simulated real small simulated scatac-seq

• A unique malignant cell type per patient tumor encoded in each cancer cell transcriptome.

  PMID 41884004 · PMC13010111 · iScience · 2026 · 8 claims · 8 setups

  Full-text index only

  Malignant cells cluster predominantly by tumor of origin, while non-malignant cells from the same tumors cluster by cell type independent of patient

  malignant non-malignant ovarian normal scrna-seq alternative paired