Research map / Probabilistic and scientific modeling / Molecules and proteins
Modeling Temporal scRNA-seq Data with Latent Gaussian Process and Optimal Transport
Mehmet Yigit Balik, Harri Lähdesmäki
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| use GP-modeled heteroscedastic noise instead of fixed noise ς(t)=0.1 (ablation) | LGP-OT · DR/Medium t=10 | Wasserstein distance W2: 37.41 -> 36.51 | Improved |
| use GP-modeled heteroscedastic noise instead of fixed noise ς(t)=0.1 (ablation) | LGP-OT · DR/Easy t=8 | Wasserstein distance W2: 31.90 -> 30.74 | Improved |
| add cell-specific latent time (ablation) | LGP-OT · DR/Medium t=10 | Wasserstein distance W2: 36.51 -> 36.33 | Improved |
| add cell type conditioning (ablation) | LGP-OT · DR/Hard t=2 | Wasserstein distance W2: 29.40 -> 28.75 | Improved |
| add cell type conditioning (ablation) | LGP-OT · DR/Easy t=8 | Wasserstein distance W2: 30.74 -> 30.59 | No clear effect |
| add both latent time and cell type conditioning (ablation) | LGP-OT · DR/Hard t=10 | Wasserstein distance W2: 38.17 -> 37.88 | No clear effect |
| use LGP-OT instead of strongest baseline | LGP-OT (w. time) · SC Hard t=17 and t=18 | Wasserstein distance W2: 17.26 -> 15.05 (t=17); 18.14 -> 15.53 (t=18) | Improved |
| use LGP-OT instead of MIOFlow | LGP-OT · DR/Hard t=10 | Wasserstein distance W2: 41.43 -> 38.17 | Improved |
| use LGP-OT instead of PRESCIENT in interpolation | LGP-OT · DR/Easy t=8 | Wasserstein distance W2: 32.48 -> 30.74 | Improved |
| perturb more lineage-specific DE genes via gradient-based trajectory optimization | LGP-OT · ZB perturbation, terminal cell type fractions | fraction of cells classified as target type: classifier identified 0.93 ±0.01 Hindbrain and 0.92 ±0.02 PSM | Improved |
| perturb non-DE genes (ablation) | LGP-OT · ZB perturbation | fraction of cells classified as target type | No clear effect |
Closest papers by meaning
- Inferring stochastic dynamics with growth from cross-sectional data (NeurIPS 2025 · similarity 0.87)
- GENOT: Entropic (Gromov) Wasserstein Flow Matching with Applications to Single-Cell Genomics (NeurIPS 2024 · similarity 0.87)
- TracingFlow: A Simulation-Free Trajectory Inference Framework Based on Second-Order Dynamics (NeurIPS 2026 · similarity 0.86)
- Learning stochastic dynamics from snapshots through regularized unbalanced optimal transport (ICLR 2025 · similarity 0.86)
- Unsupervised Gene-Cell Collective Representation Learning with Optimal Transport (AAAI 2024 · similarity 0.85)
- Enforcing Latent Euclidean Geometry in Single-Cell VAEs for Manifold Interpolation (ICML 2025 · similarity 0.85)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
