Research map / Probabilistic and scientific modeling / Uncertainty and Gaussian processes
A generative nonparametric Bayesian model for whole genomes
Alan Nawzad Amin, Eli N Weinstein, Debora Susan Marks
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 |
|---|---|---|---|
| embed linear AR model in BEAR | linear AR · YSD1 | heldout perplexity: 3.953 -> 1.144 | Improved |
| embed CNN AR model in BEAR | CNN AR · A. th. 1 | heldout perplexity: 3.947 -> 1.432 | Improved |
| embed reference-genome AR model in BEAR | reference AR · A. th. 1 | heldout perplexity: 2.686 -> 1.411 | Improved |
| embed reference-genome AR model in BEAR | reference AR · A. th. 3 | heldout perplexity: 2.340 -> 1.733 | Improved |
| embed AR model in BEAR on bacterial genomes | linear AR · Bact. | heldout perplexity: 3.831 -> 3.774 | Improved |
| embed AR model in BEAR | AR models · 10 of 11 datasets | nucleotide prediction accuracy: 30-35% -> 78-95% | Improved |
| embed linear AR model into vanilla BEAR (vs vanilla Jeffreys prior) (ablation) | BEAR · YSD1 | heldout perplexity: 1.165 -> 1.144 | Improved |
| embed reference model instead of vanilla prior (ablation) | BEAR · A. th. 1 | heldout perplexity: 1.567 -> 1.411 | Improved |
| use CNN instead of linear AR embedding (ablation) | BEAR · A. th. 1 | heldout perplexity: 1.432 -> 1.432 | No clear effect |
| use embedded AR model vs vanilla BEAR on bacterial genomes (ablation) | BEAR · Bact. | heldout perplexity: 3.774 -> 3.774 | No clear effect |
| reduce sequencing coverage from 100x to 3x | BEAR with embedded AR model · prediction accuracy gain over vanilla BEAR | accuracy points: 0.1 -> more than 10 percentage points | Improved |
| reference model vs CNN regarding misspecification | BEAR · A. th. 3 | diagnostic h: 3.964 -> 1.598 | Improved |
Closest papers by meaning
- Absorb & Escape: Overcoming Single Model Limitations in Generating Heterogeneous Genomic Sequences (NeurIPS 2024 · similarity 0.87)
- Hyperbolic Genome Embeddings (ICLR 2025 · similarity 0.86)
- DNABERT-2: Efficient Foundation Model and Benchmark For Multi-Species Genomes (ICLR 2024 · similarity 0.85)
- Context-Aware Regularization with Markovian Integration for Attention-Based Nucleotide Analysis (NeurIPS 2025 · similarity 0.85)
- Kernel-Based Evaluation of Conditional Biological Sequence Models (ICML 2024 · similarity 0.85)
- Curriculum Learning for Biological Sequence Prediction: The Case of De Novo Peptide Sequencing (ICML 2025 · similarity 0.84)
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.
