Kai J. Kohlhoff

Cloud-based simulations on Google Exacycle reveal ligand modulation of GPCR activation pathways

Simulations can provide tremendous insight into atomistic details of biological mechanisms, but micro- to milliseconds timescales are historically only accessible on dedicated supercomputers. We demonstrate that cloud computing is a viable alternative, bringing long-timescale processes within reach of a broader community. We used Googles Exacycle cloud computing platform to simulate 2 milliseconds of dynamics of the β2 adrenergic receptor — a major drug target G protein-coupled receptor (GPCR). Markov state models aggregate independent simulations into a single statistical model that is validated by previous computational and experimental results. Moreover, our models provide an atomistic description of the activation of a GPCR, revealing multiple activation pathways. Agonists and inverse agonists interact differentially with these pathways, with profound implications for drug design

Fast and accurate predictions of protein NMR chemical shifts from interatomic distances.

We present a method, CamShift, for the rapid and accurate prediction of NMR chemical shifts from protein structures. The calculations performed by CamShift are based on an approximate expression of the chemical shifts in terms of polynomial functions of interatomic distances. Since these functions are very fast to compute and readily differentiable, the CamShift approach can be utilized in standard protein structure calculation protocols.

Using NMR Chemical Shifts as Structural Restraints in Molecular Dynamics Simulations of Proteins

We introduce a procedure to determine the structures of proteins by incorporating NMR chemical shifts as structural restraints in molecular dynamics simulations. In this approach, the chemical shifts are expressed as differentiable functions of the atomic coordinates and used to compute forces to generate trajectories that lead to the reduction of the differences between experimental and calculated chemical shifts. We show that this strategy enables the folding of a set of proteins with representative topologies starting from partially denatured initial conformations without the use of additional experimental information. This method also enables the straightforward combination of chemical shifts with other standard NMR restraints, including those derived from NOE, J-coupling, and residual dipolar coupling measurements. We illustrate this aspect by calculating the structure of a transiently populated excited state conformation from chemical shift and residual dipolar coupling data measured by relaxation dispersion NMR experiments.

Dex-Net 1.0: A cloud-based network of 3D objects for robust grasp planning using a Multi-Armed Bandit model with correlated rewards

This paper presents the Dexterity Network (Dex-Net) 1.0, a dataset of 3D object models and a sampling-based planning algorithm to explore how Cloud Robotics can be used for robust grasp planning. The algorithm uses a Multi- Armed Bandit model with correlated rewards to leverage prior grasps and 3D object models in a growing dataset that currently includes over 10,000 unique 3D object models and 2.5 million parallel-jaw grasps. Each grasp includes an estimate of the probability of force closure under uncertainty in object and gripper pose and friction. Dex-Net 1.0 uses Multi-View Convolutional Neural Networks (MV-CNNs), a new deep learning method for 3D object classification, to provide a similarity metric between objects, and the Google Cloud Platform to simultaneously run up to 1,500 virtual cores, reducing experiment runtime by up to three orders of magnitude. Experiments suggest that correlated bandit techniques can use a cloud-based network of object models to significantly reduce the number of samples required for robust grasp planning. We report on system sensitivity to variations in similarity metrics and in uncertainty in pose and friction. Code and updated information is available at http://berkeleyautomation.github.io/dex-net/.

Detection of early locomotor abnormalities in a Drosophila model of Alzheimer's disease

Research highlights ► We present a computer vision system to monitor locomotion in flies in 3D. ► A Drosophila model of Alzheimers disease has been explored. ► A fast and quantitative assessment of phenotype severity is possible. ► The approach can be widely applied to different disease models.

K-Means for Parallel Architectures Using All-Prefix-Sum Sorting and Updating Steps

We present an implementation of parallel K-means clustering, called Kps-means, that achieves high performance with near-full occupancy compute kernels without imposing limits on the number of dimensions and data points permitted as input, thus combining flexibility with high degrees of parallelism and efficiency. As a key element to performance improvement, we introduce parallel sorting as data preprocessing and updating steps. Our final implementation for Nvidia GPUs achieves speedups of up to 200-fold over CPU reference code and of up to three orders of magnitude when compared with popular numerical software packages.

Science in the Cloud: Accelerating Discovery in the 21st Century

Scientific discovery is transitioning from a focus on data collection to an emphasis on analysis and prediction using large-scale computation. With appropriate software support, scientists can do these computations with unused cycles in commercial clouds. Moving science into the cloud will promote data sharing and collaborations that will accelerate scientific discovery.

The iFly tracking system for an automated locomotor and behavioural analysis of Drosophila melanogaster

The use of animal models in medical research provides insights into molecular and cellular mechanisms of human disease, and helps identify and test novel therapeutic strategies. Drosophila melanogaster--the common fruit fly--is one of the most well-established model organisms, as its study can be performed more readily and with far less expense than for other model animal systems, such as mice, fish, or primates. In the case of fruit flies, standard assays are based on the analysis of longevity and basic locomotor functions. Here we present the iFly tracking system, which enables to increase the amount of quantitative information that can be extracted from these studies, and to reduce significantly the duration and costs associated with them. The iFly system uses a single camera to simultaneously track the trajectories of up to 20 individual flies with about 100 μm spatial and 33 ms temporal resolution. The statistical analysis of fly movements recorded with such accuracy makes it possible to perform a rapid and fully automated quantitative analysis of locomotor changes in response to a range of different stimuli. We anticipate that the iFly method will reduce very considerably the costs and the duration of the testing of genetic and pharmacological interventions in Drosophila models, including an earlier detection of behavioural changes and a large increase in throughput compared to current longevity and locomotor assays.

A data-driven large-scale optimization approach for task-specific physics realism in real-time robotics simulation

Physics-based simulation of robots requires models of the simulated robots and their environment. For a realistic simulation behavior, these models must be accurate. Their physical properties such as geometric and kinematic values, as well as dynamic parameters such as mass, inertia matrix and friction, must be modelled. Unfortunately, this problem is hard for at least two reasons. First, physics engines designed for simulation of rigid bodies in real-time cannot accurately describe many common real world phenomena, e.g. (drive) friction and grasping. Second, the prime candidate solution to the model parameter problem, classical parameter identification algorithms, although well-studied and efficient, often necessitate a significant manual engineering effort and may not be applicable due to application constraints. Thus, we present a data-driven general purpose tool, which allows to optimize model parameters for (task-specific) realistic simulation behavior. Our approach directly uses the simulator and the model under optimization to improve model parameters. The optimization process is highly distributed and uses a hybrid optimization approach based on metaheuristics and the Ceres non-linear least squares solver. The user only has to provide a configuration file that specifies which model parameter to optimize together with realism criteria and a set of reference recordings from the real robot system.

Corrigendum: Cloud-based simulations on Google Exacycle reveal ligand modulation of GPCR activation pathways.

Nature Chemistry 6, 15–21 (2014); published online 15 December 2013; corrected after print 24 July 2015. In the version of this Article originally published, Figure 4 displayed incorrectly drawn chemical structures for five of the ligands. The correct structures were, however, used in the calculations.