A5045006433- Computational Drug Discovery Methods
- Protein Structure and Dynamics
- Innovative Microfluidic and Catalytic Techniques Innovation
- Parallel Computing and Optimization Techniques
- Machine Learning in Materials Science
- Protein Degradation and Inhibitors
- Scientific Computing and Data Management
- Embedded Systems Design Techniques
- Genetics, Bioinformatics, and Biomedical Research
- Advanced Neural Network Applications
- Data Management and Algorithms
- Advanced Biosensing Techniques and Applications
- Interconnection Networks and Systems
- Cloud Computing and Resource Management
- Bioinformatics and Genomic Networks
- Statistical Methods in Clinical Trials
- Electronic and Structural Properties of Oxides
- Real-Time Systems Scheduling
- Adversarial Robustness in Machine Learning
- Machine Learning and Data Classification
- Semiconductor materials and devices
- Vehicular Ad Hoc Networks (VANETs)
- Software Engineering and Design Patterns
- Transportation Safety and Impact Analysis
- Advanced Computing and Algorithms
Politecnico di Milano
2017-2024
CSC - IT Center for Science (Finland)
2023-2024
The social and economic impact of the COVID-19 pandemic demands a reduction time required to find therapeutic cure. In this paper, we describe EXSCALATE molecular docking platform capable scale on an entire modern supercomputer for supporting extreme-scale virtual screening campaigns. Such experiments can provide in short information which molecules consider next stages drug discovery pipeline, it is key asset case pandemic. has been designed benefit from heterogeneous computation nodes...
Finding a novel drug is very long and complex procedure. Using computer simulations, it possible to accelerate the preliminary phases by performing virtual screening that filters large set of candidates manageable number. This paper presents implementations comparative analysis two GPU-optimised algorithm targeting GPU architectures. work focuses on parallel computation patterns their mapping onto target architecture. The first method adopts traditional approach spreads for single molecule...
In the autonomic computing context, system is perceived as a set of autonomous elements capable self-management, where end-users define high-level goals and shall adapt to achieve desired behaviour. Runtime adaptation creates several optimization opportunities, especially if we consider approximate applications, it possible trade off accuracy result performance. Given that modern systems are limited by power dissipated, an appealing approach increase computation efficiency. this paper,...
Collaborative computing has attracted great interest in the possibility of joining efforts researchers worldwide. Its relevance further increased during pandemic crisis since it allows for strengthening scientific collaborations while avoiding physical interactions. Thus, E4C consortium presents MEDIATE initiative which invited to contribute via their virtual screening simulations that will be combined with AI-based consensus approaches provide robust and method-independent predictions. The...
In recent years, there has been a growing interest in developing high-performance implementations of drug discovery processing software. To target modern GPU architectures, such applications are mostly written proprietary languages as CUDA or HIP. However, with the increasing heterogeneity HPC systems and availability accelerators from multiple hardware vendors, it become critical to be able efficiently execute pipelines on large-scale computing systems, ultimate goal working urgent...
Incorporating speed probability distribution to the computation of route planning in car navigation systems guarantees more accurate and precise responses. In this paper, we propose a novel approach for selecting dynamically number samples used Monte Carlo simulation solve Probabilistic Time-Dependent Routing (PTDR) problem, thus improving efficiency. The proposed method is determine proactive manner simulations be done extract travel-time estimation each specific request, while respecting...
High-throughput molecular docking is a data-driven simulation methodology to estimate millions of molecules' position and interaction strength (ligands) when interacting with given protein site. Because its nature, the high-throughput performance depends on how fast we can ingest data into processing pipeline efficiently write results shared file. This work characterizes I/O high-performance, application, called Docker-HT, running supercomputer up 512 computing nodes two different parallel...
Virtual screening is one of the early stages that aims to select a set promising ligands from vast chemical library. Molecular Docking crucial task in process drug discovery and it consists estimation position molecule inside docking site. In contest urgent computing, we designed scratch EXSCALATE molecular platform benefit heterogeneous computation nodes avoid scaling issues. This poster presents achievements ongoing development platform, together with an example usage context COVID-19 pandemic.
Today digital revolution is having a dramatic impact on the pharmaceutical industry and entire healthcare system. The implementation of machine learning, extreme-scale computer simulations, big data analytics in drug design development process offers an excellent opportunity to lower risk investment reduce time patient. Within LIGATE project, we aim integrate, extend, co-design best-in-class European components Computer-Aided Drug Design (CADD) solutions exploiting today's high-end...
In a drug discovery process, the Molecular Docking task aims at estimating three-dimensional pose of molecule when it interacts with target protein. This is usually used to perform screening on large library molecules find most promising candidates. The output this estimate actual strength atomic interactions. document we focus an application that performs molecular docking using geometrical features and protein, quickly screen chemical library.
The outcome of the drug discovery process is a molecule that has strong interaction with target protein. Domain experts expect beneficial effect from this interaction. virtual screening one early stages and it aims at finding promising molecules to forward later stages. We perform task in-silico evaluate very large chemical library in short time frame. This activity typically comprises two compute-intensive tasks: docking function predicts displacement atoms, scoring function, which...
The multiplication of large integers represents a significant computation effort in some cryptographic techniques. use dedicated hardware is an appealing solution to improve performances or efficiency. We propose methodology generate throughput oriented accelerators for leveraging High-Level Synthesis. proposed micro-architectural template combines Karatsuba and Comba algorithms control the extra-functional properties generated multiplier. goal enable end user explore wide range...
Effective adaptation to climate change requires detailed, tangible information about its impacts. This paper presents a novel simulation framework that combines storyline approaches based on spectral nudging with emerging kilometre-scale global modelling capabilities using the IFS-FESOM model. approach allows for reconstructing recent extreme events and their potential evolution under different conditions while maintaining high-resolution local details. We demonstrate system's ability...
Virtual screening is an early stage of the drug discovery process that selects most promising candidates. In urgent computing scenario it critical to find a solution in short time frame. this paper, we focus on real-world virtual application evaluate out-of-kernel optimizations, consider input and architecture features improve computation efficiency GPU. Experiment results modern supercomputer node show can almost double performance. Moreover, implemented optimization using SYCL provides...
The social and economic impact of the COVID-19 pandemic demands reduction time required to find a therapeutic cure. In contest urgent computing, we re-designed Exscalate molecular docking platform benefit from heterogeneous computation nodes avoid scaling issues. We deployed on two top European supercomputers (CINECA-Marconi100 ENI-HPC5), with combined computational power 81 PFLOPS, evaluate interaction between 70 billions small molecules 15 binding-sites 12 viral proteins Sars-Cov2....
Significant improvements have been made to support the design of mixed-critical systems by developing predictable computing platforms and mechanisms for temporal spatial segregation between applications different criticalities sharing same platform. However such Multi-Processor System-on-Chips (MPSoCs) supporting needs methodologies tools improve analyzability regarding system configuration application mapping.
COVID-19 has shown the importance of having a fast response against pandemics. Finding novel drug is very long and complex procedure, it possible to accelerate preliminary phases by using computer simulations. In particular, virtual screening an in-silico phase that needed filter large set candidates manageable number. This paper presents implementations comparative analysis two GPU-optimized algorithm targeting GPU architectures. The first adopts traditional approach spreads computation...