Fabian Fröhlich

This paper introduces GEMOT, a framework for mechanistic world modeling in biology, which autonomously constructs models that match or exceed reference models in fit and parsimony across diverse biological problems.

This paper uses symbolic regression to discover compact models of intracellular signaling dynamics, showing it can recover known kinetics and identify interpretable rate laws in cancer-relevant contexts.

This paper introduces deep mechanistic models to analyze mammary MAPK signaling heterogeneity, showing that most variation comes from pathway-extrinsic factors rather than core pathway components.

This paper presents a framework for integrating single-cell and bulk data using nonlinear mixed-effect models and tailored parameter estimation, improving parameter identifiability and prediction accuracy.

The paper develops a framework to integrate multiplexed immunofluorescence data with mathematical models to study cell cycle dynamics and arrest, validated on MCF-10A cells exposed to arresting agents.

PEtab.jl improves dynamic modelling efficiency by comparing ODE solvers and parameter estimation methods, showing faster performance than pyPESTO for medium-sized models.

This study identifies a reversible basal-like cell state in AKT-inhibitor resistant triple negative breast cancer using multimodal profiling and lineage tracing.

The paper uses single-cell imaging to show that drug-tolerant cells in EGFR-mutant lung cancer rewired ERK signaling through endocytic pathways and adhesion complexes, driven by YAP and PTPRS.

The paper models MAPK signaling in BRAFV600E melanoma cells to show how allostery and rewiring lead to drug resistance.

The paper introduces a new adjoint method for computing gradients in ODE models of biochemical reaction networks using steady-state data, achieving up to 4.4x speedup in simulation time for large-scale models.

The paper investigates how varying ERK activity levels influence proliferation in cells undergoing oncogene-induced senescence, finding intermediate ERK activity promotes maximal proliferation.

The paper introduces fides, a Python framework for ODE model calibration using trust-region optimization, showing improved reliability and efficiency over existing methods.

This paper uses a pan-cancer pathway model to identify novel combination therapies and optimize treatment sequences for cancer cell lines.

The paper introduces a hierarchical method for efficiently parameterizing large-scale dynamic models using relative measurements, improving optimizer performance and enabling unbiased weighting of heterogeneous data.

This paper integrates single-cell perturbation experiments using nonlinear mixed effect modeling to robustly reconstruct cell-to-cell variability and improve parameter estimates.

The paper presents a computational framework to parameterize large-scale mechanistic models for predicting drug response in cancer cell lines using exome and transcriptome data, achieving higher accuracy than statistical approaches.

The paper benchmarks optimization methods for parameter estimation in large kinetic models and finds that a hybrid metaheuristic outperforms multi-start local searches.

This study introduces a hierarchical population model to identify latent causes of cell-to-cell variability in heterogeneous populations using single-cell data.

The paper introduces second order adjoint sensitivity analysis and a hybrid optimization-integration method to improve computational efficiency and robustness of parameter estimation for ODE models.

This paper introduces a scalable method for parameter estimation in genome-scale biochemical reaction networks using adjoint sensitivity analysis, showing improved efficiency and scalability compared to existing methods.