Research map / Probabilistic and scientific modeling / Uncertainty and Gaussian processes
Bayesian Structural Learning with Parametric Marginals for Count Data: An Application to Microbiota Systems
Veronica Vinciotti, Pariya Behrouzi, Reza Mohammadi
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 |
|---|---|---|---|
| adjust for covariates in the marginals (DWGM instead of empirical-marginal GCGM) (ablation) | Gaussian copula graphical model · Simulation with covariate effect theta1=2 | estimated graph density at 0.5 cutoff (%): 14% -> 9% | Improved |
| adjust for covariates in the marginals (DWGM instead of GCGM) (ablation) | Gaussian copula graphical model · Simulation with covariate effect theta1=2 | false positive rate (%): 9% -> 3% | Improved |
| adjust for covariates in the marginals (DWGM instead of GCGM) (ablation) | Gaussian copula graphical model · Simulation with covariate effect theta1=2 | false negative rate (%): 66% -> 67% | No clear effect |
| parametric DW marginals instead of empirical marginals, no covariate (theta1=0) (ablation) | DWGM · Simulation without covariates | AUC and KL | Mixed |
| increase number of variables p from 50 to 100 | DWGM · Simulation, n=100 | AUC and KL: p=50 -> p=100 | Hurt |
| increase graph link inclusion probability from 0.05 to 0.2 | DWGM · Simulation, n=100, p=50 | AUC and KL: 0.05 -> 0.2 | Hurt |
| increase p from 50 to 100 | DWGM · Simulation runtime | runtime (seconds): 80 -> 333 | Hurt |
| use DWGM with Bayesian edge posteriors instead of rMAGMA penalised NB-marginal copula with NB-simulated data | graph and precision estimation · Simulation with negative Binomial marginals | AUC and KL | Improved |
| add covariate adjustment (DWGM versus GCGM) on microbiome data (ablation) | Gaussian copula graphical model · HMP stool/saliva OTUs | posterior graph size | Improved |
| choose zero-inflated or standard marginal by BIC rather than always zero-inflated (ablation) | discrete Weibull marginals · HMP stool/saliva OTUs | OTUs needing zero inflation: 24 of 155 (DW), 15 of 155 (NB) | Improved |
| discrete Weibull marginals instead of negative Binomial | marginal fit of OTU counts · HMP stool/saliva OTUs | BIC: BIC difference 68.26235 vs -31.07967 | Mixed |
| set graph prior edge probability to 0.04 instead of 0.2 (ablation) | DWGM · HMP stool/saliva OTUs | correlation of edge posterior probabilities: correlation 0.97 | No clear effect |
Closest papers by meaning
- Robust and Scalable Models of Microbiome Dynamics (ICML 2018 · similarity 0.89)
- Variational Inference for sparse network reconstruction from count data (ICML 2019 · similarity 0.88)
- iLoRA: Bayesian Low-Rank Adaptation with Latent Interaction Graphs for Microbiome Diagnosis (ICML 2026 · similarity 0.86)
- Modelling Microbial Communities with Graph Neural Networks (ICML 2024 · similarity 0.85)
- A Generic Family of Graphical Models: Diversity, Efficiency, and Heterogeneity (ICML 2025 · similarity 0.85)
- Global-Local Dirichlet Processes for Clustering Grouped Data in the Presence of Group-Specific Idiosyncratic Variables (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.
