Intelligent Environments require perception systems capable of adapting to evolving tasks, heterogeneous sensing conditions, and the resource constraints of large-scale IoT deployments. This work shows an edge-centric pipeline in which a highcapacity model, specifically YOLO12n, operates at the gateway to automatically label visual data and guide the specialization of TinyML models deployed on low-power devices. Focusing on two representative mobility-related classes, cat (present in COCO) and scooter (absent from COCO), we analyze the behavior of YOLO models under zero-shot conditions, during finetuning, and when incrementally integrating new classes. Results show that YOLO12n consistently outperforms YOLO11n in zeroshot evaluation and reaches near-perfect detection accuracy (mAP@50 up to 0.995) after only a few epochs of fine-tuning, with inference times below 5 ms on GPU-equipped edge nodes. When adding the new scooter class, the model rapidly adapts, yet subsequent fine-tuning on cats reveals strong catastrophic forgetting. Rehearsal-based retraining effectively mitigates this degradation, even when using replay buffers as small as$\mathbf{1 0 - 2 0 o r i g i n a l}$dataset. These findings demonstrate that lightweight edge auto-labeling combined with efficient replay mechanisms enables sustainable, privacy-preserving, and continuously adaptive perception pipelines for next-generation Intelligent Environments and IoT monitoring infrastructures.

Monitoring invasive species is essential for protecting biodiversity in sensitive ecosystems. In the natural park of Torrevieja (Alicante, Spain), domestic cats threaten local bird populations, including endangered species. To address this issue, we developed a system leveraging deep-learning models, including YOLO (You Only Look Once), to detect cats in real-time automatically. Our solution involves deploying edge-AI cameras equipped with LoRa communication technology to efficiently transmit detection data to the cloud. This infrastructure enables continuous monitoring, accurate detection, and prompt invasive species reporting while optimizing power consumption and network bandwidth. In this paper, we present the development and evaluation of multiple deep-learning models, assess their prediction accuracy, and discuss the integration of LoRa technology to enhance data transmission in remote areas. Our findings demonstrate the feasibility and effectiveness of using edge-AI and IoT technologies for biodiversity conservation in protected natural environments.

Evapotranspiration (ET0)—the sum of evaporation and plant transpiration—is a key variable for optimizing water use in precision agriculture. With increasing challenges due to climate change and water scarcity, accurate ET0forecasting is essential for designing efficient irrigation systems that enhance productivity while conserving resources. This study evaluates advanced time-series models for ET0forecasting—Nixtla TimeGPT-1, Long Short-Term Memory Networks (LSTM), and Kolmogorov–Arnold Networks (KAN)—using IoT data from Campo de Cartagena (Murcia, Spain). Results show that KAN achieves superior performance for multi-step forecasting (MSE: 0.045), while Nixtla Linear excels in one-step predictions (MSE: 0.009). These findings provide practical insights into model selection for adaptive irrigation strategies under diverse climatic conditions.

Monitoring avian species is fundamental to detect any negative impact from human activity. In our case study, we focus in particular on the La Mata lagoon in Torrevieja (Alicante, Spain), aiming at the monitoring of two gull species that often share the same environment: Larus Michahellis (Yellow-Legged Gull), and Ichthyaetus Audouinii (Audouin’s Gull). As of 2020, Ichtyaetus Audouinii has been included within the International Union for Conservation of Nature (IUCN) Red List, with the status of vulnerable as the global population has experienced a rapid decrease, which is expected to be approaching 40% between 2006–2030, and is projected to continue declining at a similar rate over the next three generations. Since conservation of these species requires informed, prompt and efficient decision-making, we propose constantly monitoring the ecosystem by deploying an AI-enabled IoT infrastructure to listen to bird sounds, and automatically identify bird species. To this end, we relied on the BirdNET artificial neural network for the acoustic analysis. However, using BirdNET models must be carefully planned to produce insightful data that can drive informed decisions. For this reason, we discuss our methodology and the logic behind tuning the most important parameters according to the proposed use case.

Measuring ephemeral stream flows is essential for ecological and hydrological studies. However, their intermittent nature and remote locations pose challenges for conventional monitoring methods, which often consume excessive energy to capture rare events. We address this with BODOQUE (Bimodal Observational Device for Optimizing Quantification of Ephemeral streams), a dual-mode system that leverages Tiny Machine Learning (TinyML) on low-power microcontrollers. The system remains in an energy-saving sensing state and activates high-precision measurements only when water flow is detected. We present a model selection methodology that balances detection accuracy with inference cost, enabling reliable operation within hardware constraints. To enhance adaptability in diverse environments, we developed a specialized component that facilitates dataset expansion through new field samples. This supports ongoing retraining to maintain model performance under changing conditions. A comprehensive evaluation using real-world data demonstrates that our system can achieve up to 97% annual energy savings compared to traditional continuous monitoring approaches.

This paper explores the use of social sensing to analyze tourism dynamics in Torrevieja (Alicante), a tourist coastal city in Spain. By leveraging advancements in Natural Language Processing (NLP) and Deep Learning (DL), this study classifies Twitter data into ten thematic areas outlined in Torrevieja’s Tourism Development Strategy. We aim to develop a tourism-social barometer that allows real-time monitoring of significant issues impacting tourist cities, as identified through user-generated content. This analysis provides in-depth and real-time insights into public perceptions of Torrevieja as a tourist destination, factors affecting visitor experiences, and subsequent implications for destination management and marketing strategies. Employing an interdisciplinary approach that integrates tourism studies, social media analytics, and data science, our findings reveal critical themes and patterns emerging from the social media discourse. These insights enhance understanding of local tourism dynamics and also inform broader destination management practices.

Efficient long-range communication is critical for environmental monitoring, especially when dealing with large data transfers in remote areas. We present an AI-driven dynamic RF configuration mode, which combines the advanced capabilities of a novel AI*LoRa model with an enhanced RF configuration request mechanism. By leveraging TinyML, AI*LoRa dynamically adjusts key parameters of the physical layer, based on real-time environmental data, ensuring robust and energy-efficient communication. We conducted extensive real-world testing across distances ranging from a few meters to over 100 kilometers to collect the dataset necessary for training our model. The results demonstrate that our approach achieved over 90% accuracy in predicting optimal settings, leading to an average improvement of over 170% in communication efficiency. These findings underscore AI*LoRa's significant potential to enhance long-range IoT deployments.

Environmental intelligence aims to improve the decision-making process for high social and environmental value ecosystems. To this end, data are collected using different sensors to allow monitoring of different variables of interest. Typically, these ecosystems cover a large geographical area, with spots of low or no connectivity, preventing their monitoring in real time. In this work, we propose AlLoRa (Advanced Layer LoRa), a modular, low-power, long-range communication protocol based on LoRa, that allows monitoring of remote natural areas. AlLoRa has been evaluated and tested in an operational oceanographic buoy that has been deployed to address the specific environmental crisis of the Mar Menor lagoon in southeastern Spain - a region spanning 135 Km2 currently undergoing severe eutrophication process. Our results reveal that AlLoRa offers good performance regarding transfer time, power consumption, and range. The throughput ranged from around 2 kbps with SF7 to approximately 300 bps with SF11; the power consumption per kilobyte transmitted varied from 395μWh to 428μWh depending on the specific device used. The Mesh mode test successfully maintained communication between nodes over 20.33 km. Further tests in various configurations under challenging conditions validated the mesh forwarding approach. Despite tripling the distance, the system maintained reliable data transfer, improving speeds from the original point-to-point setup.

Sensors are increasingly collecting data everywhere, changing how we relate to and manage the environment. These data answer scientific questions of researchers, but can also generate social, cultural, and political effects, reinforcing the need for data in near real time. In this work, a low-power, scalable, and sustainable communication solution based on LoRa is proposed to develop an oceanographic monitoring system composed of water quality monitoring buoys that can measure, among others, parameters such as water temperature, chlorophyll-a, turbidity, oxygen concentration, etc. An exhaustive evaluation is carried out with a real test bed in a 135 Km2coastal lagoon, demonstrating that our proposal offers good performance in terms of transfer time and latency, energy consumption, and range, showing good scalability with the number of buoys added.

Tourism is one of the most relevant socio-economic sectors worldwide. However, intensive tourism has caused significant social, urban, and environmental problems. In order to improve tourism management processes and within the context of a smart tourism scenario, renewed management approaches are emerging with the aim to use the latest IT technologies to increase profits and offer new sustainable models in tourism destinations. Importantly, one key issue in tourism destinations for supporting management and planning is predicting tourist occupancy. Unfortunately, the so-called second-home tourism destinations have no reliable accommodation data coming from hospitality establishments. To overcome this pitfall, in this article, the prediction of tourist occupancy is presented based on the analysis of residential accommodation booking data and people’s comments on social networks. The analysis focuses on Torrevieja (South-eastern Spain); one of the most important second-home tourist destinations worldwide. On one hand, an ARIMA model is carried out with the time series of AirBnB bookings. On the other hand, Twitter data related to Torrevieja is analyzed by identifying main topics and entities. Our results show that AirBnB bookings estimation can be made by measuring the number of people sending posts on Twitter about tourism-related topics.

Greenhouses are complex systems where many variables are involved in order to optimize crops in an intensive agriculture framework. Therefore, monitoring and visualization of all these variables in real-time is mandatory to meet the trade-off between natural resource consumption and production maximization. In this article, we introduce an intelligent warning system to efficiently control agricultural activity in an operational greenhouse to increase productivity by optimizing crop production and energy consumption. The system includes a web application that allows the graphical and statistical representation of data measured by several sensors located inside a greenhouse. These sensors are located in strategic points that allow the reading of real-time data in a more accurate manner, therefore allowing the generation of information with the minimum percentage of error. In addition, the web application offers different data representations to allow a more exhaustive analysis of the data obtained. As a result, this warning system may help greenhouse managers to anticipate abnormal situations affecting their crops.

Effective environmental monitoring is crucial for managing global environmental challenges and providing the necessary data for Environmental Intelligence (EI). This discipline involves the integration of data from various sources to gain a comprehensive understanding of specific regions or processes. In this paper, we introduce BODOQUE, a hardware-software infrastructure to monitor water flows in ephemeral streams where the water rarely flows with great force. BODOQUE uses a low-power TinyML-based camera to detect the presence of water, activating a more complex system to measure flow only when the water flows, thereby optimizing energy consumption. This device is being deployed in the Segura basin, Murcia, Spain. This region is grappling with severe environmental issues that affect the Mar Menor, a unique saltwater lagoon. This paper focuses on the power-saving capabilities of BODOQUE, comparing the energy consumption of different edge devices running the code that measures water flow in the streams. Our goal is to determine the optimal hardware setup for the system based on our experiments, which involve performance and energy consumption tests. The results provide valuable information for future environmental monitoring systems, considering the best balance among the device's cost, performance, and energy consumption.

Abstract We are witnessing the digitalization era, where artificial intelligence (AI)/machine learning (ML) models are mandatory to transform this data deluge into actionable information. However, these models require large, high-quality datasets to predict high reliability/accuracy. Even with the maturity of Internet of Things (IoT) systems, there are still numerous scenarios where there is not enough quantity and quality of data to successfully develop AI/ML-based applications that can meet market expectations. One such scenario is precision agriculture, where operational data generation is costly and unreliable due to the extreme and remote conditions of numerous crops. In this paper, we investigated the generation of synthetic data as a method to improve predictions of AI/ML models in precision agriculture. We used generative adversarial networks (GANs) to generate synthetic temperature data for a greenhouse located in Murcia (Spain). The results reveal that the use of synthetic data significantly improves the accuracy of the AI/ML models targeted compared to using only ground truth data.

In the last decade, the popularity of UAVs has increased tremendously. Nowadays, many researchers are interested in UAV swarms. Coordinating a swarm of UAVs is a complicated task and many problems should be addressed before wide-spread adoption. In this work, we focus on the take-off for large-scale UAV swarms, with an extra focus on the assignment phase. The assignment phase is the first take-off stage whereby we decide which UAV on the ground goes to which place in the air. A good assignment algorithm, is quick, and at the same time reduce the total distance travelled as much as possible. We assess the performance of three different assignment algorithms: a heuristic, the original Kuhn-Munkres algorithm (KMA), and the KMA adapted for GPU use. Each algorithm was tested while varying the number of UAVs, as well as the type of flight formation. During the experiments, we measured the calculation time, total distance travelled, and number of flight paths crossing. In terms of total distance travelled, the KMA always outperforms the heuristic. However, the KMA takes longer (orders of magnitude) to calculate the assignment. Realistically, the KMA algorithm can only be used as long as the swarm does not contain more than 500 UAVs. From that point the GPU version of the KMA is faster. We can conclude that, in most cases, it is recommendable to use the KMA for the assignment as it will reduce the distance travelled to a minimum and, consequently, also reduce the number of flight paths crossing.

This paper focuses on the use of machine learning (ML) algorithms for greenhouses temperature forecasting. We use a new approach called TinyML, an emerging technique with high potential for a low-power environments. Our focus in this work was to understand which are the limits of this approach and how much complexity can the ML models have with the used devices. Our results reveal that a TinyML-based device such as an Arduino Nano 33 BLE Sense can execute easily low computational constrained ML models while consuming very low power compared to a high-end device. In particular, a simple multilayer perceptron (MLP) is targeted, offering accurate predictions in terms of RMSE, MAE and R2, which is remarkable considering that the models under study are extremely light and simple.

Nowadays, human overpopulation is stressing our ecosystems in different ways, being agriculture a critical example as different predictions point towards food shortages in the near future. In such context, smart farming is becoming key to optimize natural resources so that different crops are grown efficiently, consuming as few resources as possible. In particular, greenhouses have shown to be an effective approach to producing a high volume of vegetables/fruits in a reduced space and within a short time span. Hence, optimizing greenhouse functioning results in less water and nutrient consumption, less energy use, faster growth, and better product quality. In this paper, we take a step in this direction by studying the best approach to forecast greenhouse temperature based on univariate time-series analysis. In particular, several widely used time-series libraries such as Prophet by Facebook, Greykite by LinkedIn and TPOT are studied to figure out which performs better for this particular scenario. Results show that the maximum prediction error ranges from 1.5 to 3 degrees Celsius, and, in general terms, Greykite is found to be the best performing library for this particular environment.

Climate change is one of the greatest challenges for modern societies. Its consequences, often associated with extreme events, have dramatic results worldwide. New synergies between different disciplines, including artificial intelligence (AI), Internet of Things (IoT), and edge computing can lead to radically new approaches for the real-time tracking of natural disasters that are also designed to reduce the environmental footprint. In this article, we propose an AI-based pipeline for processing natural disaster images taken from drones. The purpose of this pipeline is to reduce the number of images to be processed by the first responders of the natural disaster. It consists of three main stages: 1) a lightweight autoencoder based on deep learning; 2) a dimensionality reduction using the$t$-distributed stochastic neighbor embedding algorithm; and 3) a fuzzy clustering procedure. This pipeline is evaluated on several edge computing platforms with low-power accelerators to assess the design of intelligent autonomous drones to provide this service in real time. Our experimental evaluation focuses on flooding, showing that the amount of information to be processed is substantially reduced, whereas edge computing platforms with low-power graphics accelerators are placed as a compelling alternative for processing these heavy computational workloads, obtaining a performance loss of only$2.3\times $compared to its cloud counterpart version, running both the training and inference steps.

The use of UAVs has been growing steadily over the last years. Now that even the industry is adopting them for a wide range of activities, it can be said with certainty that UAVs will become an important asset for many enterprises. We foresee that, due to affordable prices, applications with groups of UAVs, also called swarms, will become mainstream. Swarms of UAVs can perform tasks faster and/or with more redundancy, and other tasks are only possible by collaborative work of UAVs. However, there are still many challenges to be solved before swarms of UAVs can be used safely. One of the challenges is the takeoff; i.e., takeoff should be safe (no collisions) and fast at the same time. An important part of the takeoff is the assignment task; i.e., determining which UAV goes where. In this work we will compare the effectiveness of three assignment algorithms, in terms of total distance travelled, number of flight paths crossing, and calculation time. We specially focus on irregular patterns. Our results show that the Kuhn-Munkres Algorithm (KMA) is preferable in almost all cases. It ensures that the total distance travelled by all UAVs is minimal, and most importantly it reduces the number of flight paths crossing each other (i.e. potential collisions). This is a very important metric because it allows for fast (semi) simultaneous takeoff procedures, which are not possible if the chances of collision are high.

The field of Unmanned Aerial Vehicles (UAVs) is gaining momentum thanks to the amazing capabilities of these flying devices. In particular, small aircrafts using vertical takeoff and landing (VTOL) are among the preferred solutions in the civilian sector thanks to their low cost, simplicity of operation, and the ability to carry powerful sensing devices. When combined to create a swarm, the potential of such UAVs is further extended by allowing to perform more complex missions efficiently. However, as the number of UAVs involved becomes higher, many issues arise that can result into mission failures. In this paper, we specifically address the swarm takeoff problem from an optimization perspective. We propose a new takeoff scheme based on the Munkres algorithm that solves the assignment problem in polynomial time. Our evaluation studies the taking off complexity of large swarms and analyze the computational and quality trade-off of our proposal. Experiments show that the Munkres algorithm offers optimal solution with a low computation overhead.

MOTIVATION: Molecular docking methods are extensively used to predict the interaction between protein-ligand systems in terms of structure and binding affinity, through the optimization of a physics-based scoring function. However, the computational requirements of these simulations grow exponentially with: (i) the global optimization procedure, (ii) the number and degrees of freedom of molecular conformations generated and (iii) the mathematical complexity of the scoring function. RESULTS: In this work, we introduce a novel molecular docking method named METADOCK 2, which incorporates several novel features, such as (i) a ligand-dependent blind docking approach that exhaustively scans the whole protein surface to detect novel allosteric sites, (ii) an optimization method to enable the use of a wide branch of metaheuristics and (iii) a heterogeneous implementation based on multicore CPUs and multiple graphics processing units. Two representative scoring functions implemented in METADOCK 2 are extensively evaluated in terms of computational performance and accuracy using several benchmarks (such as the well-known DUD) against AutoDock 4.2 and AutoDock Vina. Results place METADOCK 2 as an efficient and accurate docking methodology able to deal with complex systems where computational demands are staggering and which outperforms both AutoDock Vina and AutoDock 4. AVAILABILITY AND IMPLEMENTATION: https://[email protected]/Baldoimbernon/metadock_2.git. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Summary Agriculture is one of the key sectors where technology is opening new opportunities to break up the market. The Internet of Things (IoT) could reduce the production costs and increase the product quality by providing intelligence services via IoT analytics. However, the hard weather conditions and the lack of connectivity in this field limit the successful deployment of such services as they require both, ie, fully connected infrastructures and highly computational resources. Edge computing has emerged as a solution to bring computing power in close proximity to the sensors, providing energy savings, highly responsive web services, and the ability to mask transient cloud outages. In this paper, we propose an IoT monitoring system to activate anti‐frost techniques to avoid crop loss, by defining two intelligent services to detect outliers caused by the sensor errors. The former is a nearest neighbor technique and the latter is the k‐means algorithm, which provides better quality results but it increases the computational cost. Cloud versus edge computing approaches are analyzed by targeting two different low‐power GPUs. Our experimental results show that cloud‐based approaches provides highest performance in general but edge computing is a compelling alternative to mask transient cloud outages and provide highly responsive data analytic services in technologically hostile environments.

Summary Virtual Screening (VS) methods simulate molecular interactions in silico to look for the best chemical compound that interacts with a given molecular target. VS is becoming increasingly popular to accelerate the drug discovery process and constitute hard optimization problems with a huge computational cost. To deal with these two challenges, we have created METADOCK, an application that (1) enables a wide range of metaheuristics through a parametrized schema and (2) promotes the use of a multi‐GPU environment within a heterogeneous cluster. Metaheuristics provide approximate solutions in a reasonable time frame, but, given the stochastic nature of real‐life procedures, the energy budget goes hand in hand with acceleration to validate the proposed solution. This paper evaluates energy trade‐offs and correlations with performance for a set of metaheuristics derived from METADOCK. We establish a solid inference from minimal power to maximal performance in GPUs, and from there, to optimal energy consumption. This way, ideal heuristics can be chosen according not only to best accuracy and performance but also to energy requirements. Our study starts with a preselection of parameterized metaheuristic functions, building blocks where we will find optimal patterns from power criteria while preserving parallelism through a GPU execution. We then establish a methodology to figure out the best instances of the parameterized kernels based on energy patterns obtained, which are analyzed from different viewpoints, ie, performance, average power, and total energy consumed. We also compare the best workload distributions for optimal performance and power efficiency among Pascal and Maxwell GPUs on popular Titan models. Our experimental results demonstrate that the most power efficient GPU can be overloaded in order to reduce the total amount of energy required by as much as 20%, finding unique scenarios where Maxwell does it better in execution time, but with Pascal always ahead in performance per watt, reaching peaks of up to 40%.

Novel cooperative intelligent transportation systems (ITS) serve as the basis for the provision of a number of services for drivers, occupants, and third parties. The vast amount of information to be collected, especially in vehicle-to-infrastructure (V2I) communication services, requires new algorithms and hardware platforms to cope with real-time requirements; however, this combination is not properly addressed in the literature. In this paper, we introduce a high-throughput hardware-software infrastructure to gather information from vehicles and efficiently process it to provide novel ITS services. We propose a parallelization approach of a fuzzy clustering technique on heterogeneous servers based on CPU and several GPUs, tailored to classification problems in V2I. The infrastructure is empirically tested to offer a geo-located pollution information service through the periodical collection of both vehicle's position and status data. We offer a real service that correctly identifies highly polluting traffic areas and drivers. The results indicate a good performance of the system under high loads, and our scalability analysis reveals a good operation in real-ambitious deployments thanks to the use of the both CPU and multiple GPUs, showing that our proposal can efficiently host cooperative services involving high processing in the ITS context.

The efficient market hypothesis (EMH) affirms that asset prices should reveal all available information. Therefore, it is impossible to "beat the market" always on a risk-adjusted basis since market prices should only respond to new information. Here, we propose a new model to validate the EMH that is inspired on an elastic network model. More specifically, we apply this comparison to Foreign Exchange (FOREX) market under some restrictive conditions. In our hypothesis, several interaction potentials are used to characterize the interaction between banks and each particular quotation. This hypothesis comes from the study of several natural systems, such as macromolecules in solution. An algorithm based on the Monte Carlo methods is also presented in order to predict the evolution of the system.

ABSTRACT The integration of the latest breakthroughs in computational modeling and high performance computing (HPC) has leveraged advances in the fields of healthcare and drug discovery, among others. By integrating all these developments together, scientists are creating new exciting personal therapeutic strategies for living longer that were unimaginable not that long ago. However, we are witnessing the biggest revolution in HPC in the last decade. Several graphics processing unit architectures have established their niche in the HPC arena but at the expense of an excessive power and heat. A solution for this important problem is based on heterogeneity. In this paper, we analyze power consumption on heterogeneous systems, benchmarking a bioinformatics kernel within the framework of virtual screening methods. Cores and frequencies are tuned to further improve the performance or energy efficiency on those architectures. Our experimental results show that targeted low‐cost systems are the lowest power consumption platforms, although the most energy efficient platform and the best suited for performance improvement is the Kepler GK110 graphics processing unit from Nvidia by using compute unified device architecture. Finally, the open computing language version of virtual screening shows a remarkable performance penalty compared with its compute unified device architecture counterpart. Copyright © 2013 John Wiley & Sons, Ltd.

SUMMARY Virtual Screening (VS) methods can considerably aid drug discovery research, predicting how ligands interact with drug targets. BINDSURF is an efficient and fast blind VS methodology for the determination of protein binding sites, depending on the ligand, using the massively parallel architecture of graphics processing units(GPUs) for fast unbiased prescreening of large ligand databases. In this contribution, we provide a performance/cost model for the execution of this application on both local system and public cloud infrastructures. With our model, it is possible to determine which is the best infrastructure to use in terms of execution time and costs for any given problem to be solved by BINDSURF. Conclusions obtained from our study can be extrapolated to other GPU‐based VS methodologies.Copyright © 2013 John Wiley & Sons, Ltd.

Virtual Screening (VS) methods can considerably aid clinical research, predicting how ligands interact with drug targets. Most VS methods suppose a unique binding site for the target, but it has been demonstrated that diverse ligands interact with unrelated parts of the target and many VS methods do not take into account this relevant fact. This problem is circumvented by a novel VS methodology named BINDSURF that scans the whole protein surface to find new hotspots, where ligands might potentially interact with, and which is implemented in massively parallel Graphics Processing Units, allowing fast processing of large ligand databases. BINDSURF can thus be used in drug discovery, drug design, drug repurposing and therefore helps considerably in clinical research. However, the accuracy of most VS methods is constrained by limitations in the scoring function that describes biomolecular interactions, and even nowadays these uncertainties are not completely understood. In order to solve this problem, we propose a novel approach where neural networks are trained with databases of known active (drugs) and inactive compounds, and later used to improve VS predictions.

The current trend in medical research for the discovery of new drugs is the use of Virtual Screening (VS) methods. In these methods, the calculation of the non-bonded interactions, such as electrostatics or van der Waals forces, plays an important role, representing up to 80% of the total execution time. These kernels are computational intensive and massively parallel in nature, and thus they are well suited to be accelerated on parallel architectures. In this work, we discuss the effective parallelization of the non-bonded electrostatic interactions kernel for VS on three different parallel architectures: a shared memory system, a distributed memory system, and a Graphics Processing Units (GPUs). For an efficient handling of the computational intensive and massively parallelism of this kernel, we enable different data policies on those architectures to take advantage of all computational resources offered by them. Four implementations are provided based on MPI, OpenMP, Hybrid MPI Open MP and CUDA programming models. The sequential implementation is defeated by a wide margin by all parallel implementations, obtaining up to 72x speed-up factor on the shared memory system through OpenMP, up to 60x and229x speed-ups factors on the distributed memory system for the MPI implementation and the Hybrid MPI-Open MP implementation respectively, and finally, up to 213x speedup factor for the CUDA implementation on the GPU architecture to offer the best alternative in terms of performance/cost ratio.

Diffusion Weighted Magnetic Resonance Imaging (DW-MRI) and tractography approaches are the only tools that can be utilized to estimate structural connections between different brain areas, non-invasively and in-vivo. A first step that is commonly utilized in these techniques includes the estimation of the underlying fibre orientations and their uncertainty in each voxel of the image. A popular method to achieve that is implemented in the FSL software, provided by the FMRIB Centre at University of Oxford, and is based on a Bayesian inference framework. Despite its popularity, the approach has high computational demands, taking normally more than 24 hours for analyzing a single subject. In this paper, we present a GPU-optimized version of the FSL tool that estimates fibre orientations. We report up to 85x of speed-up factor between the GPU and its sequential counterpart CPU-based version.

BACKGROUND: Virtual Screening (VS) methods can considerably aid clinical research, predicting how ligands interact with drug targets. Most VS methods suppose a unique binding site for the target, usually derived from the interpretation of the protein crystal structure. However, it has been demonstrated that in many cases, diverse ligands interact with unrelated parts of the target and many VS methods do not take into account this relevant fact. RESULTS: We present BINDSURF, a novel VS methodology that scans the whole protein surface in order to find new hotspots, where ligands might potentially interact with, and which is implemented in last generation massively parallel GPU hardware, allowing fast processing of large ligand databases. CONCLUSIONS: BINDSURF is an efficient and fast blind methodology for the determination of protein binding sites depending on the ligand, that uses the massively parallel architecture of GPUs for fast pre-screening of large ligand databases. Its results can also guide posterior application of more detailed VS methods in concrete binding sites of proteins, and its utilization can aid in drug discovery, design, repurposing and therefore help considerably in clinical research.

P systems or Membrane Systems provide a high-level computational modelling framework that combines the structure and dynamic aspects of biological systems in a relevant and understandable way. They are inherently parallel and non-deterministic computing devices. In this article, we discuss the motivation, design principles and key of the implementation of a simulator for the class of recognizer P systems with active membranes running on a (GPU). We compare our parallel simulator for GPUs to the simulator developed for a single central processing unit (CPU), showing that GPUs are better suited than CPUs to simulate P systems due to their highly parallel nature.