Process Simulation of Complex Biological Pathways in Physical Reactive Space and Reformulated for Massively Parallel Computing Platforms.

Biological systems encompass complexity that far surpasses many artificial systems. Modeling and simulation of large and complex biochemical pathways is a computationally intensive challenge. Traditional tools, such as ordinary differential equations, partial differential equations, stochastic master equations, and Gillespie type methods, are all limited either by their modeling fidelity or computational efficiency or both. In this work, we present a scalable computational framework based on modeling biochemical reactions in explicit 3D space, that is suitable for studying the behavior of large and complex biological pathways. The framework is designed to exploit parallelism and scalability offered by commodity massively parallel processors such as the graphics processing units (GPUs) and other parallel computing platforms. The reaction modeling in 3D space is aimed at enhancing the realism of the model compared to traditional modeling tools and framework. We introduce the Parallel Select algorithm that is key to breaking the sequential bottleneck limiting the performance of most other tools designed to study biochemical interactions. The algorithm is designed to be computationally tractable, handle hundreds of interacting chemical species and millions of independent agents by considering all-particle interactions within the system. We also present an implementation of the framework on the popular graphics processing units and apply it to the simulation study of JAK-STAT Signal Transduction Pathway. The computational framework will offer a deeper insight into various biological processes within the cell and help us observe key events as they unfold in space and time. This will advance the current state-of-the-art in simulation study of large scale biological systems and also enable the realistic simulation study of macro-biological cultures, where inter-cellular interactions are prevalent.

CUDAMPF: a multi-tiered parallel framework for accelerating protein sequence search in HMMER on CUDA-enabled GPU.

BACKGROUND: HMMER software suite is widely used for analysis of homologous protein and nucleotide sequences with high sensitivity. The latest version of hmmsearch in HMMER 3.x, utilizes heuristic-pipeline which consists of MSV/SSV (Multiple/Single ungapped Segment Viterbi) stage, P7Viterbi stage and the Forward scoring stage to accelerate homology detection. Since the latest version is highly optimized for performance on modern multi-core CPUs with SSE capabilities, only a few acceleration attempts report speedup. However, the most compute intensive tasks within the pipeline (viz., MSV/SSV and P7Viterbi stages) still stand to benefit from the computational capabilities of massively parallel processors. RESULTS: A Multi-Tiered Parallel Framework (CUDAMPF) implemented on CUDA-enabled GPUs presented here, offers a finer-grained parallelism for MSV/SSV and Viterbi algorithms. We couple SIMT (Single Instruction Multiple Threads) mechanism with SIMD (Single Instructions Multiple Data) video instructions with warp-synchronism to achieve high-throughput processing and eliminate thread idling. We also propose a hardware-aware optimal allocation scheme of scarce resources like on-chip memory and caches in order to boost performance and scalability of CUDAMPF. In addition, runtime compilation via NVRTC available with CUDA 7.0 is incorporated into the presented framework that not only helps unroll innermost loop to yield upto 2 to 3-fold speedup than static compilation but also enables dynamic loading and switching of kernels depending on the query model size, in order to achieve optimal performance. CONCLUSIONS: CUDAMPF is designed as a hardware-aware parallel framework for accelerating computational hotspots within the hmmsearch pipeline as well as other sequence alignment applications. It achieves significant speedup by exploiting hierarchical parallelism on single GPU and takes full advantage of limited resources based on their own performance features. In addition to exceeding performance of other acceleration attempts, comprehensive evaluations against high-end CPUs (Intel i5, i7 and Xeon) shows that CUDAMPF yields upto 440 GCUPS for SSV, 277 GCUPS for MSV and 14.3 GCUPS for P7Viterbi all with 100 % accuracy, which translates to a maximum speedup of 37.5, 23.1 and 11.6-fold for MSV, SSV and P7Viterbi respectively. The source code is available at https://github.com/Super-Hippo/CUDAMPF.

Synthesis of optimal control of complex biological pathways enabled by global sensitivity analysis.

The past few years have witnessed growth in the number of system biological models corresponding to several different pathways and have shed light on biological processes governing vital functions such as signal transduction, cell-proliferation and cell-apoptosis. Among several challenges in modeling and verification, significant efforts have been made in system identification including model parameter estimation and key component identification. In a practical or experimental setting, the effects of various control strategies, are usually associated with costs including enzyme consumption or the desired levels of output transcription. In this work we address the problem of designing an optimal control law via a combination of computational tools such as sensitivity analysis, Pontryagin maximum principle and steepest descent technique. We find that, in the presence of limitation of enzyme concentration level control, optimal control law can be developed only by choosing enzyme with dominant effect on dynamical behavior of biological pathway. In this work, we study JAK/STAT signal transduction pathway, and by simulation analysis, we investigate the effectiveness of Sobol values in developing optimal control law.

Structural, dynamic, and electrostatic properties of fully hydrated DMPC bilayers from molecular dynamics simulations accelerated with graphical processing units (GPUs).

We present results of molecular dynamics simulations of fully hydrated DMPC bilayers performed on graphics processing units (GPUs) using current state-of-the-art non-polarizable force fields and a local GPU-enabled molecular dynamics code named FEN ZI. We treat the conditionally convergent electrostatic interaction energy exactly using the particle mesh Ewald method (PME) for solution of Poisson's Equation for the electrostatic potential under periodic boundary conditions. We discuss elements of our implementation of the PME algorithm on GPUs as well as pertinent performance issues. We proceed to show results of simulations of extended lipid bilayer systems using our program, FEN ZI. We performed simulations of DMPC bilayer systems consisting of 17,004, 68,484, and 273,936 atoms in explicit solvent. We present bilayer structural properties (atomic number densities, electron density profiles), deuterium order parameters (S(CD)), electrostatic properties (dipole potential, water dipole moments), and orientational properties of water. Predicted properties demonstrate excellent agreement with experiment and previous all-atom molecular dynamics simulations. We observe no statistically significant differences in calculated structural or electrostatic properties for different system sizes, suggesting the small bilayer simulations (less than 100 lipid molecules) provide equivalent representation of structural and electrostatic properties associated with significantly larger systems (over 1000 lipid molecules). We stress that the three system size representations will have differences in other properties such as surface capillary wave dynamics or surface tension related effects that are not probed in the current study. The latter properties are inherently dependent on system size. This contribution suggests the suitability of applying emerging GPU technologies to studies of an important class of biological environments, that of lipid bilayers and their associated integral membrane proteins. We envision that this technology will push the boundaries of fully atomic-resolution modeling of these biological systems, thus enabling unprecedented exploration of meso-scale phenomena (mechanisms, kinetics, energetics) with atomic detail at commodity hardware prices.

Computational approaches to spatial orientation: from transfer functions to dynamic Bayesian inference.

Spatial orientation is the sense of body orientation and self-motion relative to the stationary environment, fundamental to normal waking behavior and control of everyday motor actions including eye movements, postural control, and locomotion. The brain achieves spatial orientation by integrating visual, vestibular, and somatosensory signals. Over the past years, considerable progress has been made toward understanding how these signals are processed by the brain using multiple computational approaches that include frequency domain analysis, the concept of internal models, observer theory, Bayesian theory, and Kalman filtering. Here we put these approaches in context by examining the specific questions that can be addressed by each technique and some of the scientific insights that have resulted. We conclude with a recent application of particle filtering, a probabilistic simulation technique that aims to generate the most likely state estimates by incorporating internal models of sensor dynamics and physical laws and noise associated with sensory processing as well as prior knowledge or experience. In this framework, priors for low angular velocity and linear acceleration can explain the phenomena of velocity storage and frequency segregation, both of which have been modeled previously using arbitrary low-pass filtering. How Kalman and particle filters may be implemented by the brain is an emerging field. Unlike past neurophysiological research that has aimed to characterize mean responses of single neurons, investigations of dynamic Bayesian inference should attempt to characterize population activities that constitute probabilistic representations of sensory and prior information.