Georg Kurt Gerber

MCSPACE: inferring microbiome spatiotemporal dynamics from high-throughput co-localization data

MCSPACE infers microbiome spatiotemporal dynamics from co-localization data, showing superior performance over existing methods in mice and human datasets.

MMETHANE: interpretable AI for predicting host status from microbial composition and metabolomics data

MMETHANE predicts host status from microbial and metabolomics data using an interpretable deep learning model, outperforming existing methods on most datasets and uncovering biologically meaningful links.

Microbial dynamics inference at ecosystem-scale

The paper introduces MDSINE2, a Bayesian method for inferring microbial dynamics from time-series data, and shows it outperforms existing methods in identifying ecosystem-scale interactions.

Strain Tracking with Uncertainty Quantification

ChronoStrain tracks low-abundance microbial strains over time with uncertainty quantification, outperforming existing methods on real and synthetic data.

Gut Microbiome Signatures During Acute Infection Predict Long COVID

The paper identifies distinct gut microbiome signatures during acute SARS-CoV-2 infection that predict long COVID, showing microbiome profiles better predict risk than clinical variables.

AI-driven Discovery of Morphomolecular Signatures in Toxicology

This paper introduces GEESE, an AI model that imputes morphomolecular signatures in toxicology data, linking gene expression changes to drug-induced liver injuries.

MDITRE: scalable and interpretable machine learning for predicting host status from temporal microbiome dynamics

MDITRE predicts host status from microbiome time-series data using interpretable deep learning, outperforming existing methods in speed and accuracy.

Scalable learning of interpretable rules for the dynamic microbiome domain

The paper introduces a scalable, differentiable model for learning interpretable rules from microbiome time-series data to classify host status, achieving comparable performance to a Bayesian method with faster runtime and more rules.

MITRE: predicting host status from microbiota time-series data

MITRE predicts host status from microbiome time-series data, performing on par or better than black-box methods while maintaining interpretability.

Engineered inter-species amino acid cross-feeding increases population evenness in a synthetic bacterial consortium

The paper engineers inter-species amino acid cross-feeding in bacterial consortia to increase population evenness in synthetic microbial communities.

A synthetic bacterial information transfer system functions in the mammalian gut

The paper engineers a synthetic bacterial communication system in the mammalian gut using quorum sensing and bacteriophage genetic switch circuits.