Research map / Graph learning / Graph self-supervised learning
Metagenomic Binning using Connectivity-constrained Variational Autoencoders
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Add assembly-graph edge loss (VAE+E) to VAE training (ablation) | VAMB* (VAE) · Real-world datasets (6 WWTP) | total HQ bins: 256.3 -> 265.8 | Helped |
| Add SCG loss on top of edge loss (VAE+E+SCG) (ablation) | VAE+E · Real-world datasets (6 WWTP) | total HQ bins: 265.8 -> 266.2 | Helped |
| Add assembly-graph edge loss (VAE+E) to VAE training (ablation) | VAMB* (VAE) · STRONG100 (simulated) | F1: 0.878 -> 0.887 | Helped |
| Add SCG loss on top of edge loss (ablation) | VAE+E · STRONG100 (simulated) | F1: 0.887 -> 0.903 | Helped |
| Full CCVAE (edge + SCG losses) vs MetaBAT2 baseline | CCVAE · Real-world datasets (6 WWTP) | total HQ bins: 253 -> 266.2 | Helped |
| Full CCVAE vs best baseline on simulated data | CCVAE · STRONG100 (simulated) | F1: 0.870 (VAMB) -> 0.903 | Helped |
| Add SCG loss and edge loss on AALE dataset (ablation) | VAMB* (VAE) · AALE | HQ bins: 55.8 -> 60.4 | Helped |
| Add edge and SCG losses on MARI, HJOR, VIBY and DAMH | CCVAE · DAMH / HJOR / VIBY (real) | HQ bins: DAMH 49.8 vs 49.8 (VAMB*); HJOR 27.8 vs 26.2; VIBY 29.0 vs 27.6 | No clear effect |
Closest papers by meaning
- RepBin: Constraint-Based Graph Representation Learning for Metagenomic Binning (AAAI 2022 · similarity 0.89)
- Encoding Unitig-level Assembly Graphs with Heterophilous Constraints for Metagenomic Contigs Binning (ICLR 2024 · similarity 0.89)
- Autoencoding Random Forests (NeurIPS 2025 · similarity 0.83)
- Context-Aware Regularization with Markovian Integration for Attention-Based Nucleotide Analysis (NeurIPS 2025 · similarity 0.83)
- PhyloVAE: Unsupervised Learning of Phylogenetic Trees via Variational Autoencoders (ICLR 2025 · similarity 0.83)
- C3P: Contrastive promoter-protein pretraining yields representations capturing bacterial gene regulation (NeurIPS 2026 · similarity 0.82)
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.