Gerhard Hummer · Goethe University Frankfurt · h-index 111
Hummer's group develops and applies computational methods to understand biomolecular structure, dynamics, and function across multiple scales. The program spans from fundamental biophysics—including water transport through carbon nanotubes [1,2,3] and molecular dynamics simulations with periodic boundary conditions [4]—to biomedically relevant systems like SARS-CoV-2 viral proteins [10]. Recent work emphasizes integrating computational modeling with experimental data to characterize intrinsically disordered proteins and protein assemblies [7,9]. The group also investigates cellular processes including ubiquitination-regulated ER-phagy and membrane remodeling [11,12], and inflammatory signaling pathways [1].
The group specializes in molecular dynamics simulations and free energy calculations, particularly developing methods to extract thermodynamic and kinetic information from non-equilibrium single-molecule pulling experiments [1,2]. A key methodological focus is Bayesian ensemble refinement by reweighting, which efficiently combines computational ensembles with experimental data (SAXS, FRET, NMR) by optimizing log-weights rather than forces—achieving ~20-fold speedup for large ensembles (N=10^6) [5,6,8]. The group also performs coarse-grained molecular dynamics to access nanosecond timescales from sub-picosecond simulations [13] and atomistic simulations of membrane systems including pore formation [14].
Hummer's early work demonstrated single-file water conduction through hydrophobic carbon nanotube channels [1,2] and osmotic water transport through carbon nanotube membranes [3]. His group developed optimized force fields for peptide helix-coil transitions [5] and methods to extract intrinsic rates and activation free energies from single-molecule experiments [2]. Recent structural biology contributions include revealing flexibility in SARS-CoV-2 spike protein mediated by three hinges [8] and demonstrating how the papain-like protease regulates viral spread and innate immunity [10]. Work on intrinsically disordered proteins showed that tau's global structure emerges from local structure [7] and that disease-linked TDP-43 hyperphosphorylation suppresses condensation [9].
The program is advancing integrative structural biology approaches that combine multiple experimental techniques with computational ensemble refinement [5,6,7]. There is continued emphasis on understanding cellular mechanisms, particularly ubiquitin-regulated processes in ER remodeling and selective autophagy [11,12,15]. The group is also exploring membrane-associated phenomena including gasdermin-D pore formation in atomistic detail [14] and lipid regulation mechanisms [3]. The methodological trajectory points toward increasingly efficient algorithms for large-scale ensemble optimization that enable exploring different subensembles and experimental datasets with varying confidence levels [5].
Free energy reconstruction from single-molecule pulling experiments relies fundamentally on the Jarzynski equality, but requires important extensions to handle the specific challenges of force spectroscopy.
The core difficulty is that while "Jarzynski's remarkable identity" provides a connection between nonequilibrium work and free energy differences, "it relates the nonequilibrium work to free energy differences at different times, not positions" [2]. This is problematic because pulling experiments naturally measure forces and extensions (positions), not time-dependent free energies.
Hummer and colleagues overcame this challenge by developing methods to extract equilibrium free energy profiles as a function of molecular extension. Their approach shows that "equilibrium free energy profiles can be extracted rigorously from repeated nonequilibrium force measurements on the basis of an extension of Jarzynski's remarkable identity" [3]. Specifically, "by surmounting this difficulty," they "were able to express the free energy profile in terms of the integral of the force with respect to extension" [2].
A more refined method uses an inverse Weierstrass transform approach, where "an inverse Weierstrass transform is used to relate the system free energy obtained from the Jarzynski equality directly to the underlying molecular free energy surface" [1]. This avoids the need for "work-weighted position histograms" and provides "an accurate approximation for the free energy surface...by using the method of steepest descent to evaluate the inverse transform" [1].
The key innovation is thus the mathematical framework that bridges Jarzynski's time-based equality to position-based free energy profiles relevant to force spectroscopy experiments.
Based on the provided passages, Hummer's work shows several key findings about water transport through carbon nanotubes:
1. Burst-like conduction with collective motion: Water conduction through carbon nanotube channels occurs in bursts with collective water motion rather than as independent molecular events [2].
2. Single-file arrangement: Water molecules are confined in a single-file arrangement within the nanotube channel, and transport involves concerted movement of these molecules [2].
3. Sequential filling/emptying mechanism: The kinetics of water filling and emptying occur predominantly by sequential addition or removal of water molecules to/from a single-file chain inside the nanotube [3].
4. Orientational ordering: Both advancing and receding water chains are orientationally ordered. This ordering prevents simultaneous filling from both tube ends and forces chain rupturing to occur at the tube end where a water molecule donates a hydrogen bond to the bulk fluid [3].
5. Hydrogen-bonded molecular wires: Water confined in narrow carbon nanotube channels forms collectively oriented molecular wires held together by tight hydrogen bonds [4].
The work demonstrates that water transport through carbon nanotubes exhibits highly correlated, collective behavior that can be described by continuous-time random-walk models for single-file transport [2].
Generated by LabBrain — a research-literature assistant I built:
it resolves a PI on OpenAlex, ingests their open-access full text (Europe PMC
structured XML → PDF → abstract), builds a hybrid dense + BM25 retrieval index
with reciprocal-rank fusion and LLM reranking, and answers questions grounded in the
corpus with title · year · DOI · exact-passage citations. Embedding runs on CPU
(ROCm-safe); every model call is timeout-guarded.
— Gianangelo Dichio · MS Data Science + BS Mathematics