Keyong Li

Achieving reliability and high accuracy in automated protein docking: Cluspro, PIPER, SDU, and stability analysis in CAPRI rounds 13–19

Our approach to protein—protein docking includes three main steps. First, we run PIPER, a rigid body docking program based on the Fast Fourier Transform (FFT) correlation approach, extended to use pairwise interactions potentials. Second, the 1000 best energy conformations are clustered, and the 30 largest clusters are retained for refinement. Third, the stability of the clusters is analyzed by short Monte Carlo simulations, and the structures are refined by the medium‐range optimization method SDU. The first two steps of this approach are implemented in the ClusPro 2.0 protein–protein docking server. Despite being fully automated, the last step is computationally too expensive to be included in the server. When comparing the models obtained in CAPRI rounds 13–19 by ClusPro, by the refinement of the ClusPro predictions and by all predictor groups, we arrived at three conclusions. First, for the first time in the CAPRI history, our automated ClusPro server was able to compete with the best human predictor groups. Second, selecting the top ranked models, our current protocol reliably generates high‐quality structures of protein–protein complexes from the structures of separately crystallized proteins, even in the absence of biological information, provided that there is limited backbone conformational change. Third, despite occasional successes, homology modeling requires further improvement to achieve reliable docking results. Proteins 2010.

Data-rate requirements for nonlinear Feedback control

Abstract This paper and talk describe recent research on communication and information processing requirements for implementing nonlinear feedback control laws in applications such as the coordinated motion control of groups of mobile robots. The principal focus is on how the well-known data-rate theorem might be reformulated in the context of essentially nonlinear control problems.

Robust quantization for digital finite communication bandwidth (DFCB) control

In this paper, we consider a scalar model of DFCB control that accommodates time-varying data rate constraints, such as might occur with intermittent network congestion, and asynchronism of sampling and control actuation. Because of the possibly unpredictable fluctuation of the data rate, we are interested in feedback control designs that tolerate significantly constrained data-rates on feedback loops, while providing acceptable performance when such data rate constraints are not in force. In light of a very basic notion of acceptable performance, we show that control designs with different number of quantization levels tolerate constrained data rates differently. This leads to the conclusion that binary control represents the most robust control quantization under data rate constraints imposed by time-varying congestion on the feedback communication channel. The advantage margin of binary control is further investigated numerically with and without the sampling-control asynchronism being considered. We show that the advantage margin is more substantial when the sampling-control asynchronism is significant.

Encounter complexes and dimensionality reduction in protein–protein association

An outstanding challenge has been to understand the mechanism whereby proteins associate. We report here the results of exhaustively sampling the conformational space in protein–protein association using a physics-based energy function. The agreement between experimental intermolecular paramagnetic relaxation enhancement (PRE) data and the PRE profiles calculated from the docked structures shows that the method captures both specific and non-specific encounter complexes. To explore the energy landscape in the vicinity of the native structure, the nonlinear manifold describing the relative orientation of two solid bodies is projected onto a Euclidean space in which the shape of low energy regions is studied by principal component analysis. Results show that the energy surface is canyon-like, with a smooth funnel within a two dimensional subspace capturing over 75% of the total motion. Thus, proteins tend to associate along preferred pathways, similar to sliding of a protein along DNA in the process of protein-DNA recognition. DOI: http://dx.doi.org/10.7554/eLife.01370.001

Model-free probabilistic localization of wireless sensor network nodes in indoor environments

We present a technique that makes up a practical probabilistic approach for locating wireless sensor network devices using the commonly available signal strength measurements (RSSI). From the RSSI measurements between transmitters and receivers situated on a set of landmarks, we construct appropriate probabilistic descriptors associated with a devices position in the contiguous space using a pdf interpolation technique. We then develop a localization system that relies on these descriptors and the measurements made by a set of clusterheads positioned at some of the landmarks. The localization problem is formulated as a composite hypothesis testing problem. We develop the requisite theory, characterize the probability of error, and address the problem of optimally placing clusterheads. Experimental results show that our system achieves an accuracy equivalent to 95% < 5 meters and 87% < 3 meters.

Problems in Decentralized Sensor-Actuator Networks

There is a growing body of literature on networked control systems treating the relationship between network channel capacity and stability of the system’s operation. In a very rough but intuitive sense, the main results in this area provide a quantitative understanding of the way in which restrictions on the rate of information exchange among system components in a real-time system will degrade the system’s performance. Recent extensions of these results provide an understanding of how system performance will depend on the magnitude of noise and the degree of asynchronism in the operation of system components. A number of researchers have recently begun to look at the problem constraints on feedback channel capacity in decentralized feedback control structures. In the present paper, we examine the way in which decentralization magnifies the degradation of information due to noise and asynchronism among decentralized sensors and leads to instabilities even in cases where feedback channels have ample capacity for stable operation of a system with centralized components. Further, we discuss an approach to solving the observed problem by a novel source-coding strategy which is similar to but different from the well known Gray code.

An actor-critic method using Least Squares Temporal Difference learning

In this paper, we use a Least Squares Temporal Difference (LSTD) algorithm in an actor-critic framework where the actor and the critic operate concurrently. That is, instead of learning the value function or policy gradient of a fixed policy, the critic carries out its learning on one sample path while the policy is slowly varying. Convergence of such a process has previously been proven for the first order TD algorithms, TD(λ) and TD(1). However, the conversion to the more powerful LSTD turns out not straightforward, because some conditions on the stepsize sequences must be modified for the LSTD case. We propose a solution and prove the convergence of the process. Furthermore, we apply the LSTD actor-critic to an application of intelligently dispatching forklifts in a warehouse.

Landmark-based position and movement detection of wireless sensor network devices

The demand for reliable indoor positioning of mobile devices is wide and the computing and communication hardware are evolving rapidly. Thus, it is meaningful to consider a spectrum of techniques that reflect different constraints and tradeoffs of hardware investment, computational complexity, set-up cost, and positioning accuracy. The present paper strives to achieve high accuracy with low hardware investment and moderate set-up cost, but with somewhat sophisticated computations. We not only describe a successful positioning system, but also suggest a set of formal techniques that proved to work well in the real setting, and can be implemented using standard wireless sensor network hardware. The overall philosophy of utilizing the full distribution information of signal measurements at each location (proved implementable after careful algorithm design) distinguishes sharply from most existing works. Compared with the earlier work within our group, the present paper introduces new elements such as profiling the signal measurement distributions over the coverage area using a special interpolation technique; a two-tier tracking scheme that improves the efficiency of localization (in the commonly used sense) by adding movement detection (lower energy cost); and the joint clusterhead placement optimization for both localization and movement detection. Experimentally, our system achieved an accuracy equivalent to 95% < 5 meters and 87% < 3 meters, which should be considered a high-quality result compared to well-known contemporary systems that use similar low-cost hardware.

Distribution-Dependent Robust Linear Optimization with Asymmetric Uncertainty and Application to Optimal Control

Abstract We consider a linear programming problem in which the constraint matrix is uncertain. Each element of the constraint matrix is modeled as a random variable whose range is asymmetrically bounded around its mean. We construct a formulation that yields a solution with a better objective value, compared to the classical robust optimization approach, while taking the risk that the solution may become infeasible to the original problem. We address the risk by establishing upper bounds on the probability that it violates the constraints of the problem. These bounds exploit full distributional information on the random elements or limited distributional information such as the true means or sample means of the random elements. We explore the application of our methodology to the optimal control of linear uncertain systems with constraints.

Position and Movement Detection of Wireless Sensor Network Devices Relative to a Landmark Graph

We present a novel probabilistic framework for reliable indoor positioning of mobile sensor network devices. Compared to existing approaches, ours adopts complex computations in exchange for high localization accuracy while needing low hardware investment and moderate set-up cost. To that end, we use full distributional information on signal measurements at a set of discrete locations, termed landmarks. Positioning of a mobile device is done relative to the resulting landmark graph and the device can be found near a landmark or in the area between two landmarks. Key elements of our approach include profiling the signal measurement distributions over the coverage area using a special interpolation technique; a two-tier statistical positioning scheme that improves efficiency by adding movement detection; and joint clusterhead placement optimization for both localization and movement detection. The proposed system is practical and has been implemented using standard wireless sensor network hardware. Experimentally, our system achieved an accuracy equivalent to less than 5 meters with 95 percent success probability and less than 3 meters with an 87 percent success probability. This performance is superior to well-known contemporary systems that use similar low-cost hardware.