The introduction of accelerator devices such as graphics processing units (GPUs) has had profound impact on molecular dynamics simulations and has enabled order-of-magnitude performance advances using commodity hardware. To fully reap these benefits, it has been necessary to reformulate some of the most fundamental algorithms, including the Verlet list, pair searching, and cutoffs. Here, we present the heterogeneous parallelization and acceleration design of molecular dynamics implemented in the GROMACS codebase over the last decade. The setup involves a general cluster-based approach to pair lists and non-bonded pair interactions that utilizes both GPU and central processing unit (CPU) single instruction, multiple data acceleration efficiently, including the ability to load-balance tasks between CPUs and GPUs. The algorithm work efficiency is tuned for each type of hardware, and to use accelerators more efficiently, we introduce dual pair lists with rolling pruning updates. Combined with new direct GPU–GPU communication and GPU integration, this enables excellent performance from single GPU simulations through strong scaling across multiple GPUs and efficient multi-node parallelization.

SUMMARY Fibrinogen, upon enzymatic conversion to monomeric fibrin, provides the building blocks for fibrin polymer, the scaffold of blood clots and thrombi. Little has been known about the force-induced unfolding of fibrin(ogen), even though it is the foundation for the mechanical and rheological properties of fibrin, which are essential for hemostasis. We determined mechanisms and mapped the free energy landscape of the elongation of fibrin(ogen) monomers and oligomers through combined experimental and theoretical studies of the nanomechanical properties of fibrin(ogen), using atomic force microscopy-based single-molecule unfolding and simulations in the experimentally relevant timescale. We have found that mechanical unraveling of fibrin(ogen) is determined by the combined molecular transitions that couple stepwise unfolding of the γ chain nodules and reversible extension-contraction of the α-helical coiled-coil connectors. These findings provide important characteristics of the fibrin(ogen) nanomechanics necessary to understand the molecular origins of fibrin viscoelasticity at the fiber and whole clot levels.

We characterized the α-to-β transition in α-helical coiled-coil connectors of human fibrin(ogen) molecule using biomolecular simulations of their forced elongation, and theoretical modeling. The force (F) - extension (X) profiles show three distinct regimes: (1) the elastic regime, in which the coiled-coils act as entropic springs (F < 100–125 pN; X < 7–8 nm); (2) the constant-force plastic regime, characterized by a force-plateau (F≈150 pN; X≈10–35 nm); and (3) the non-linear regime (F >175–200 pN; X > 40–50 nm). In the plastic regime, the three-stranded α-helices undergo a non-cooperative phase transition to form parallel three-stranded β-sheets. The critical extension of α-helices is 0.25 nm, and the energy difference between the α-helices and β-sheets is 4.9 kcal/mol per helical pitch. The soft α-to-β phase transition in coiled-coils might be a universal mechanism underlying mechanical properties of filamentous α-helical proteins.

Theoretical exploration of fundamental biological processes involving the forced unraveling of multimeric proteins, the sliding motion in protein fibers and the mechanical deformation of biomolecular assemblies under physiological force loads is challenging even for distributed computing systems. Using a C(α)-based coarse-grained self organized polymer (SOP) model, we implemented the Langevin simulations of proteins on graphics processing units (SOP-GPU program). We assessed the computational performance of an end-to-end application of the program, where all the steps of the algorithm are running on a GPU, by profiling the simulation time and memory usage for a number of test systems. The ∼90-fold computational speedup on a GPU, compared with an optimized central processing unit program, enabled us to follow the dynamics in the centisecond timescale, and to obtain the force-extension profiles using experimental pulling speeds (v(f) = 1-10 μm/s) employed in atomic force microscopy and in optical tweezers-based dynamic force spectroscopy. We found that the mechanical molecular response critically depends on the conditions of force application and that the kinetics and pathways for unfolding change drastically even upon a modest 10-fold increase in v(f). This implies that, to resolve accurately the free energy landscape and to relate the results of single-molecule experiments in vitro and in silico, molecular simulations should be carried out under the experimentally relevant force loads. This can be accomplished in reasonable wall-clock time for biomolecules of size as large as 10(5) residues using the SOP-GPU package.

The use of graphics processing units (GPUs) in simulation applications offers a significant speed gain as compared to computations on central processing units (CPUs). Many simulation methods require a large number of independent random variables generated at each step. We present two approaches for implementation of random number generators (RNGs) on a GPU. In the one-RNG-per-thread approach, one RNG produces a stream of random numbers in each thread of execution, whereas the one-RNG-for-all-threads method builds on the ability of different threads to communicate, thus, sharing random seeds across an entire GPU device. We used these approaches to implement Ran2, Hybrid Taus, and Lagged Fibonacci algorithms on a GPU. We profiled the performance of these generators in terms of the computational time, memory usage, and the speedup factor (CPU time/GPU time). These generators have been incorporated into the program for Langevin simulations of biomolecules fully implemented on the GPU. The ∼250-fold computational speedup on the GPU allowedus to carry out single-molecule dynamic force measurements in silico to explore the mechanical properties of the bacteriophage HK97 in the experimental subsecond time scale. We found that the nanomechanical response of HK97 depends on the conditions of force application, including the rate of change and geometry of the mechanical perturbation. Hence, using the GPU-based implementation of RNGs, presented here, in conjunction with Langevin simulations, makes it possible to directly compare the results of dynamic force measurements in vitro and in silico.

Thirteen tubulin protofilaments, made of αβ-tubulin heterodimers, interact laterally to produce cytoskeletal microtubules. Microtubules exhibit the striking property of dynamic instability, manifested in their intermittent growth and shrinkage at both ends. This behavior is key to many cellular processes, such as cell division, migration, maintenance of cell shape, etc. Although assembly and disassembly of microtubules is known to be linked to hydrolysis of a guanosine triphosphate molecule in the pocket of β-tubulin, detailed mechanistic understanding of corresponding conformational changes is still lacking. Here we take advantage of the recent generation of in-microtubule structures of tubulin to examine the properties of protofilaments, which serve as important microtubule assembly and disassembly intermediates. We find that initially straight tubulin protofilaments, relax to similar non-radially curved and slightly twisted conformations. Our analysis further suggests that guanosine triphosphate hydrolysis primarily affects the flexibility and conformation of the inter-dimer interface, without a strong impact on the shape or flexibility of αβ-heterodimer. Inter-dimer interfaces are significantly more flexible compared to intra-dimer interfaces. We argue that such a difference in flexibility could be key for distinct stability of the plus and minus microtubule ends. The higher flexibility of the inter-dimer interface may have implications for development of pulling force by curving tubulin protofilaments during microtubule disassembly, a process of major importance for chromosome motions in mitosis.

Background: Knob-hole interactions underlie formation and properties of fibrin polymer, the scaffold of blood clots and thrombi. Results: The structural mechanisms, dissociation kinetics, and thermodynamic parameters of the A:a and B:b knob-hole interactions have been determined. Conclusion: The knob-hole bonds are inherently variable and sensitive to pH and temperature. Significance: The emerging molecular picture offers mechanistic insights into fibrin polymerization.

SUMMARY Fibrin is a filamentous network made in blood to stem bleeding; it forms when fibrinogen is converted into fibrin monomers that self-associate into oligomers and then to polymers. To gather structural insights into fibrin formation and properties, we combined high-resolution atomic force microscopy of fibrin(ogen) oligomers and molecular modeling of crystal structures of fibrin(ogen) and its fragments. We provided a structural basis for the intermolecular flexibility of single-stranded fibrin(ogen) oligomers and identified a hinge region at the D:D inter-monomer junction. Following computational reconstruction of the missing portions, we recreated the full-atomic structure of double-stranded fibrin oligomers that was validated by quantitative comparison with the experimental images. We characterized previously unknown intermolecular binding contacts at the D:D and D:E:D interfaces, which drive oligomerization and reinforce the intra- and inter-strand connections in fibrin besides the known knob-hole bonds. The atomic models provide valuable insights into the submolecular mechanisms of fibrin polymerization.