This paper proposes a ski training system using VR (Virtual Reality) that enables beginners to acquire skiing skills without going to an actual ski ground. The proposed system obtains the speed of skiing based on the center of pressure (COP) of each player's foot. The first-person perspective of skiing at the obtained speed down a ski slope is fed back to the player as a VR image. Experiments were conducted to evaluate the effectiveness of the proposed system and the VR interface. Specifically, beginner skiers were categorized into three groups: “a group trained with the proposed VR system”, “a group trained with a system that provides feedback of the skiing speed calculated from the COP by increasing or decreasing the gauge (a bar-shaped graph representing changes in numerical values), instead of VR”, and “a group that does not train with the system”. After training under each of these conditions, a sliding test was conducted on an actual ski slope to check the degree of skill acquisition. The results show that subjects trained with the proposed system acquired more skiing skills than subjects who did not use the system on actual ski slopes. Furthermore, there was no clear difference in the result of the sliding test between subjects trained by the VR interface and those trained by the gauge interface, but the VR interface yields better deceleration postures.

MOTIVATION: N4-methylcytosine (4mC) is an essential kind of epigenetic modification that regulates a wide range of biological processes. However, experimental methods for detecting 4mC sites are time-consuming and labor-intensive. As an alternative, computational methods that are capable of automatically identifying 4mC with data analysis techniques become a reasonable option. A major challenge is how to develop effective methods to fully exploit the complex interactions within the DNA sequences to improve the predictive capability. RESULTS: In this work, we propose MSNet-4mC, a lightweight neural network building upon convolutional operations with multi-scale receptive fields to perceive cross-element relationships over both short and long ranges of given DNA sequences. With strong imbalances in the number of candidates in different species in mind, we compute and apply class weights in the cross-entropy loss to balance the training process. Extensive benchmarking experiments show that our method achieves a significant performance improvement and outperforms other state-of-the-art methods. AVAILABILITY AND IMPLEMENTATION: The source code and models are freely available for download at https://github.com/LIU-CT/MSNet-4mC, implemented in Python and supported on Linux and Windows. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

This paper proposes a method that can classify foot strike types using a deep learning model and can extract unknown factors, which enables to evaluate running motions without being influenced by biases of sports experts, using the contribution degree of input values (CDIV). Accelerometers are attached to the runner's body, and when the runner runs, a fixed camera observes the runner and acquires a video sequence synchronously with the accelerometers. To train a deep learning model for classifying foot strikes, we annotate foot strike acceleration data for RFS (Rearfoot strike) or non-RFS objectively by watching the video. To interpret the unknown factors extracted from the learned model, we calculate two CDIVs: the contributions of the resampling time and the accelerometer value to the output (foot strike type). Experiments on classifying unknown runners' foot strikes were conducted. As a common result to sport science, it is confirmed that the CDIVs contribute highly at the time of the right foot strike, and the sensor values corresponding to the right and left tibias contribute highly to classifying the foot strikes. Experimental results show the right tibia is important for classifying foot strikes. This is because many of the training data represent difference between the two foot strikes in the right tibia. As a conclusion, our proposed method could extract unknown factors from the classifier and could interpret the factors that contain similar knowledge to the prior knowledge of experts, as well as new findings that are not included in conventional knowledge.

SUMMARY: KofamKOALA is a web server to assign KEGG Orthologs (KOs) to protein sequences by homology search against a database of profile hidden Markov models (KOfam) with pre-computed adaptive score thresholds. KofamKOALA is faster than existing KO assignment tools with its accuracy being comparable to the best performing tools. Function annotation by KofamKOALA helps linking genes to KEGG resources such as the KEGG pathway maps and facilitates molecular network reconstruction. AVAILABILITY AND IMPLEMENTATION: KofamKOALA, KofamScan and KOfam are freely available from GenomeNet (https://www.genome.jp/tools/kofamkoala/). SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

To improve running performances, each runner’s skill, such as characteristics and habits, needs to be known, and feedback on the performance should be outputted according to the runner's skill level. In this paper, we propose a new coaching system for detecting the skill of a runner and a method of giving feedback using a sprint motion dataset. Our proposed method calculates an extracted feature to detect the skill using an autoencoder whose middle layer is an LSTM layer; we analyse the feature using hierarchical clustering, and we analyse the human joints that affect the skill. As a result of experiments, five clusters are obtained using hierarchical clustering. This paper clarifies how to detect the skill and to output feedback to achieve a level of performance one step higher than the current level.

For building a disaster response robot, WAREC-l's fully-automated system for manipulating a valve, this paper proposes a method for (1) detecting a valve which is far away from the robot, (2) estimating the position and orientation for grasping the valve by the robot at a closer position. Our methods do not need any prior information about a valve for the above-mentioned detection and estimation for grasping. In addition, our estimation for grasping provides useful information, by which WAREC-I can rotate a valve autonomously. The method (1) uses the RGB image and the point cloud data captured by Multisense SL as the input, and estimate the position and orientation of a valve far away from the robot. The method (2) uses both the RGB and depth images captured by KinectV2 as input and estimate information for grasping the valve. Our experiments are conducted using a real disaster response robot. Our experimental results show the error of the estimation by the (a) two methods are small enough to achieve a fully-automated system for detecting and rotating the valve by WAREC-I.

SUMMARY: ViPTree is a web server provided through GenomeNet to generate viral proteomic trees for classification of viruses based on genome-wide similarities. Users can upload viral genomes sequenced either by genomics or metagenomics. ViPTree generates proteomic trees for the uploaded genomes together with flexibly selected reference viral genomes. ViPTree also serves as a platform to visually investigate genomic alignments and automatically annotated gene functions for the uploaded viral genomes, thus providing virus researchers the first choice for classifying and understanding newly sequenced viral genomes. AVAILABILITY AND IMPLEMENTATION: ViPTree is freely available at: http://www.genome.jp/viptree . CONTACT: [email protected]. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

MOTIVATION: Functional prediction of paralogs is challenging in bioinformatics because of rapid functional diversification after gene duplication events combined with parallel acquisitions of similar functions by different paralogs. Plant type III polyketide synthases (PKSs), producing various secondary metabolites, represent a paralogous family that has undergone gene duplication and functional alteration. Currently, there is no computational method available for the functional prediction of type III PKSs. RESULTS: We developed a plant type III PKS reaction predictor, pPAP, based on the recently proposed classification of type III PKSs. pPAP combines two kinds of similarity measures: one calculated by profile hidden Markov models (pHMMs) built from functionally and structurally important partial sequence regions, and the other based on mutual information between residue positions. pPAP targets PKSs acting on ring-type starter substrates, and classifies their functions into four reaction types. The pHMM approach discriminated two reaction types with high accuracy (97.5%, 39/40), but its accuracy decreased when discriminating three reaction types (87.8%, 43/49). When combined with a correlation-based approach, all 49 PKSs were correctly discriminated, and pPAP was still highly accurate (91.4%, 64/70) even after adding other reaction types. These results suggest pPAP, which is based on linear discriminant analyses of similarity measures, is effective for plant type III PKS function prediction. AVAILABILITY AND IMPLEMENTATION: pPAP is freely available at ftp://ftp.genome.jp/pub/tools/ppap/. CONTACT: [email protected]. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Recent education roles encourage willingness to learn individually, solve unfamiliar problems using knowledge acquired through information and communication technologies (ICTs), and collaborate with others. Various learning methods to cultivate this ability have been invented and researchers discussed learning effect on such methods with qualitative and quantitative analyses. One of the authors introduced pair programming as a method of peer learning and flip teaching, which consists of preliminary learning of basic programming and advanced learning practices based on peer activity in classroom lessons, into computer simulation practices for undergraduate students. With introducing class schedule design for flip teaching and development of peer learning preparation support system for determining the appropriate pair formation and seat allocation in the classroom utilizing a probabilistic combinatorial optimization algorithm, this study focuses on learning behavior of wellperforming students, analyzing learning records and access logs on a learning management system and answers to a questionnaire administered after the practices. In this analysis, we attempted to discriminate behavior of well-performing students observed from the learning records, access logs, and questionnaire to discover best practices for improving the performance of medium to bottom-line students. Well-performing students tended to prepare for classroom lessons in good time, but their performance depended on the lesson content difficulty. A correlation was observed between the frequency of interaction among students and skill acquisition. We discuss how to improve learning environments using ICTs and collaborative learning methods based on the analysis.

We consider a teaching system that covers the stages of task from planning to on-site teaching, and propose for it a structured description based on task properties. A multimodal communication-teaching-advisor (MCTA) system is used for on-site teaching for a tool-manipulating task. The modular structure and object-oriented task handling program enable it to be applied to various robot systems and lets system developers customize the interface easily. The interface allows the operator and robot to share internal and external sensor data via nonverbal media. The intentions of an inexpert operator, who has only common knowledge and ordinary abilities, can then be transferred to the robot system via nonverbal operation. Experiments show that two different types of robot controller could be controlled by this teaching system, and the nominal task program was designed for an environment with two controllers.

The development of robot systems must be simplified and made less expensive so that they can be used more widely in various industrial fields. We propose a component architecture of a robot-teaching system based on a universal task model, that enables fast and easy development of robot-teaching systems for various types of work.

A model of contact-state transitions is useful for motion planning, robot control, and sensing. Modeling the entire set of contact-state transitions every object, however, is very troublesome. To overcome the problem, we have developed a method for programming the model from human demonstration using a model of local contact-state transitions between a vertex pair (local model). Our method consists of three parts: (1) allocating the local model to vertex pairs where contact-state transitions will occur, (2) gathering them into groups according to an assembly sequence, and (3) unifying different groups resulting from different demonstrations of the same task. The allocation and grouping are done by simply showing a robotic system an assembly sequence through demonstration. The unification is done by comparing local models between different groups. We applied the method to two object-insertion tasks and did two kinds of experiments: (1) programming groups of local models from demonstration and (2) performing the demonstrated task by a robot using the programmed groups. The results showed the effectiveness in programming from demonstration and in the modeling problem.

A new framework is proposed for generating trajectories for robots. This planning by analogy approach uses problems and solutions acquired from previous operator teaching or automatic path planning. Unlike traditional path-planning methods, users need not specify goals or prepare search heuristics. Instead, users specify the correspondence of objects in the task environment to those in a reference environment. Path planning by analogy reduces the amount of operator instruction and the computational cost of making a path, and can generate a trajectory close to the desired trajectory.

The ability to execute similar tasks in various environments is important for robot systems, but there are few approaches for doing this. We recently proposed a "teaching by showing" system that recognizes teaching tasks and executes corresponding tasks in different environments. The teaching task is recognized from geometrical constraint and translated into a general task description (GTD). However, this method is difficult to apply in complex environments because of redundancy an description and lack of generalization. This paper presents an algorithm for both two- and three-dimensional environments that solves these problems. The improved GTD describes tasks in complex environments very compactly. It ignores objects that do not affect the task, which adds a useful robustness to the system.>

A "teaching by showing" system using a graphic interface is proposed. This system provides: 1) a user interface with which an inexperienced operator can easily teach a task, and 2) a task execution system that can operate the task in a different environment. The operator shows an example of assembly task movements in the virtual environmental created in the computer. A finite automaton dependently defined task is used to interpret the movements as a sequence of high-level representations of operations. When the system is commanded to operate the learning task, the system observes the task environment, checks the geometrical feasibility of the task in the environment, and if necessary replans the sequence of operations so that the robot can complete the task. Then the system uses task-dependent interpretation rules to translate the sequence of operations into manipulator-level commands and executes the task by replicating the operator's movement in the virtual environment.>

The authors propose a robot teaching interface which uses virtual reality. The teaching method provides a user interface with which a novice operator can easily direct a robot. The operator performs the assembly task in a virtual workspace generated by a computer. The operator's movements are recognized as robot task-level operations by using a finite automation. The system interprets the recognized operations into manipulator-level commands using task-dependent interpretation rules and a world model. A robot executes the assembly task in the workplace by replicating the operator's movements in the virtual workspace.>