Sabry Razick

• Authen, Kristian; Riemer-Sørensen, Signe; Michałowska, Katarzyna; Vedvik, Emil; Razick, Sabry & Visoka, Kushtrim (2026). WindAI: Wind power forecasting in Norway – data competition summary. Nordic Machine Intelligence (NMI). 6(2), p. 9–9. doi: 10.5617/nmi.13106. Full text in Research Archive Show summary

  This paper describes the WindAI data challenge on short-term (48 hours ahead) wind power forecasting in Norway. The competition provided multi-year open datasets combining ensemble weather forecasts, nowcasts, aggregated wind power production, and wind park metadata for Norwegian bidding zones. Participants were tasked with producing day-ahead hourly forecasts under realistic operational constraints. We summarize the data, task design, evaluation protocol, and the modeling approaches adopted by the top-performing teams, and discuss key findings regarding model complexity, feature engineering, and robustness.

• Samarakoon, Pubudu Saneth; Fournous, Ghislain; Hansen, Lars Thomas; Wijesiri, Ashen; Zhao, Sen & A, Rodriguez Alex [Show all 12 contributors for this article] (2025). Benchmarking accelerated next-generation sequencing analysis pipelines. Bioinformatics Advances. 5(1), p. 1–13. doi: 10.1093/bioadv/vbaf085. Full text in Research Archive Show summary

  Motivation Industry-standard central processing unit (CPU)-based next-generation sequencing (NGS) analysis tools have led to longer runtimes, affecting their utility in time-sensitive clinical practices and population-scale research studies. To address this, researchers have developed accelerated NGS platforms like DRAGEN and Parabricks, which have significantly reduced runtimes—from days to hours. However, these studies have evaluated accelerated platforms independently without sufficiently assessing computational resource usage or thoroughly investigating speedup scalability, a gap our study is designed to address. Results Corroborating previous studies, accelerated pipelines demonstrated shorter runtimes than CPU-only approaches, with Parabricks-H100 demonstrating the highest speedups, followed by DRAGEN. In mapping, DRAGEN outperformed Parabricks (L4 and A100) and matched H100 speedups. Parabricks (A100 and H100) variant calling demonstrated higher speedups than DRAGEN. Moreover, DRAGEN and Parabricks-H100 mapping showed positive trends in the coverage-based scalability analysis, while other configurations failed to scale effectively. Our profiler analysis provided new insights into the relationships between Parabricks’ performances and resource usage patterns, revealing its potential for further improvements. Our findings and cost comparison help researchers select accelerated platforms based on coverage needs, timeframes, and budget, while suggesting optimization strategies.

• Turinsky, Andrei L.; Razick, Sabry; Turner, Brian G.; Donaldson, Ian M & Wodak, Shoshana J. (2014). Navigating the global protein-protein interaction landscape using iRefWeb. Methods in molecular biology. ISSN 1064-3745. 1091, p. 315–331. doi: 10.1007/978-1-62703-691-7-22. Full text in Research Archive
• Razick, Sabry; Mocnik, Rok; Thomas, Laurent; Ryeng, Einar; Drabløs, Finn & Sætrom, Pål (2014). The eGenVar data management system--cataloguing and sharing sensitive data and metadata for the life sciences. Database: The Journal of Biological Databases and Curation. doi: 10.1093/database/bau027. Full text in Research Archive Show summary

  Systematic data management and controlled data sharing aim at increasing reproducibility, reducing redundancy in work, and providing a way to efficiently locate complementing or contradicting information. One method of achieving this is collecting data in a central repository or in a location that is part of a federated system and providing interfaces to the data. However, certain data, such as data from biobanks or clinical studies, may, for legal and privacy reasons, often not be stored in public repositories. Instead, we describe a metadata cataloguing system and a software suite for reporting the presence of data from the life sciences domain. The system stores three types of metadata: file information, file provenance and data lineage, and content descriptions. Our software suite includes both graphical and command line interfaces that allow users to report and tag files with these different metadata types. Importantly, the files remain in their original locations with their existing access-control mechanisms in place, while our system provides descriptions of their contents and relationships. Our system and software suite thereby provide a common framework for cataloguing and sharing both public and private data.

• Razick, Sabry; Ortiz, Antonio Carlos Mora; Michalickova, Katerina; Boddie, Paul & Donaldson, Ian Matthew (2011). iRefScape. A Cytoscape plug-in for visualization and data mining of protein interaction data from iRefIndex. BMC Bioinformatics. 12(1), p. 388–388. doi: 10.1186/1471-2105-12-388. Full text in Research Archive Show summary

  Background The iRefIndex consolidates protein interaction data from ten databases in a rigorous manner using sequence-based hash keys. Working with consolidated interaction data comes with distinct challenges: data are redundant, overlapping, highly interconnected and may be collected and represented using different curation practices. These phenomena were quantified in our previous studies. Results The iRefScape plug-in for the Cytoscape graphical viewer addresses these challenges. We show how these factors impact on data-mining tasks and how our solutions resolve them in a simple and efficient manner. A uniform accession space is used to limit redundancy and support search expansion and searching on multiple accession types. Multiple node and edge features support data filtering and mining. Node colours and features supply information about search result provenance. Overlapping evidence is presented using a multi-graph and a bi-partite representation is used to distinguish binary and n-ary source data. Searching for interactions between sets of proteins is supported and specifically includes searches on disease-related genes found in OMIM. Finally, a synchronized adjacency-matrix view facilitates visualization of relationships between sets of user defined groups. Conclusions The iRefScape plug-in will be of interest to advanced users of interaction data. The plug-in provides access to a consolidated data set in a uniform accession space while remaining faithful to the underlying source data. Tools are provided to facilitate a range of tasks from a simple search to knowledge discovery. The plug-in uses a number of strategies that will be of interest to other plug-in developers.

• Turner, Brian; Mohammed, Razick Ahmed Sabry; Turinsky, Andrei L.; Vlasblom, James; Crowdy, Edgard K. & Cho, Emerson [Show all 9 contributors for this article] (2010). iRefWeb: interactive analysis of consolidated protein interaction data and their supporting evidence. Database: The Journal of Biological Databases and Curation. doi: 10.1093/database/baq023. Full text in Research Archive
• Turinsky, Andrei L.; Mohammed, Razick Ahmed Sabry; Turner, Brian; Donaldson, Ian Matthew & Wodak, Shoshana J. (2010). Literature curation of protein interactions: measuring agreement across major public databases. Database: The Journal of Biological Databases and Curation. doi: 10.1093/database/baq026. Full text in Research Archive
• Mohammed, Razick Ahmed Sabry; Magklaras, Georgios Vasilei & Donaldson, Ian Matthew (2008). iRefIndex: a consolidated protein interaction database with provenance. BMC Bioinformatics. 9, p. 405–405. doi: 10.1186/1471-2105-9-405. Full text in Research Archive Show summary

  Background Interaction data for a given protein may be spread across multiple databases. We set out to create a unifying index that would facilitate searching for these data and that would group together redundant interaction data while recording the methods used to perform this grouping. Results We present a method to generate a key for a protein interaction record and a key for each participant protein. These keys may be generated by anyone using only the primary sequence of the proteins, their taxonomy identifiers and the Secure Hash Algorithm. Two interaction records will have identical keys if they refer to the same set of identical protein sequences and taxonomy identifiers. We define records with identical keys as a redundant group. Our method required that we map protein database references found in interaction records to current protein sequence records. Operations performed during this mapping are described by a mapping score that may provide valuable feedback to source interaction databases on problematic references that are malformed, deprecated, ambiguous or unfound. Keys for protein participants allow for retrieval of interaction information independent of the protein references used in the original records. Conclusion We have applied our method to protein interaction records from BIND, BioGrid, DIP, HPRD, IntAct, MINT, MPact, MPPI and OPHID. The resulting interaction reference index is provided in PSI-MITAB 2.5 format at http://irefindex.uio.no. This index may form the basis of alternative redundant groupings based on gene identifiers or near sequence identity groupings.

View all works in NVA

• Hansen, Lars Thomas; Stensland, Håkon Kvale; Roehr, Thomas Mirko; Wijesiri, Ashen & Razick, Sabry (2026). WP9: Best-Fit Infrastructure. doi: 10.5281/zenodo.20699921. Full text in Research Archive Show summary

  An essential basis towards the development of a best-fit infrastructure solution in WP9 has been the development of a consistent monitoring solution for cluster loads and job efficiencies. The tool, which as a whole is called Jobanalyzer, has the following components at a high level

• Trotter, James David; Larsen, Tore Heide; Roehr, Thomas Mirko & Razick, Sabry (2026). NAIC Deliverable Report - Benchmarking D9.5. doi: 10.5281/ZENODO.19834281. Full text in Research Archive Show summary

  This report details efforts to document and empirically evaluate an existing computing infrastructure. In this context, we focused on benchmarking GPU performance for ML/AI workloads, primarily using the available hardware on eX3, an experimental computing infrastructure hosted at Simula.

• Saruar, Alam; Alexander, Oltu; Razick, Sabry; Thomas, Röblitz; Kristiansen, Gunleiksrud Daniel & Buket, Ogul Burcin (2026). Final Report on NAIC-WP4: Project-specific assistance, help desk, and advanced user support. doi: 10.5281/ZENODO.19202091. Full text in Research Archive Show summary

  This report summarizes the achievements of Work Package 4 (WP4) in establishing comprehensive user support and project-specific assistance for the Norwegian AI Cloud (NAIC). NAIC WP4 gained valuable insights about user needs and patterns, informing future national research support offerings in Norway.

• -, BURÇİN, Buket Ogul & Razick, Sabry (2026). Final Report - WP3 (Competence Building, Training and Knowledge Hub). doi: 10.5281/zenodo.19460112. Full text in Research Archive Show summary

  WP3 contributed in five main areas: 1. Competence mapping and training needs identification The work package collected and structured information about existing training, staff competence needs, and onboarding gaps for both users and support environments. 2. Training development and delivery WP3 organized, facilitated, co-hosted, or supported a range of training activities covering AI/ML, HPC, accelerated computing, scientific visualization, and domain-specific topics such as bioinformatics. 3. Pilot infrastructure for asynchronous learning A local **Moodle** instance was deployed and tested in EduCloud (moodle.uio.no) in order to assess the feasibility of a more structured, self-paced learning environment for NAIC-related training. 4. Knowledge hub and learning structure WP3 consolidated training-related material into a more structured framework, including Learning Paths and Ready-to-Integrate Modules, intended as baseline material for future reuse and expansion. 5. Groundwork for long-term continuation Several activities initiated in WP3 were completed as NAIC deliverables but are intended to continue in a more operational and national form under KI-fabrikken. WP3 has contributed not only through individual training activities, but also by helping shape a more structured approach to AI and HPC competence development at the national level, and by providing a basis for further work under KI-fabrikken.

• Razick, Sabry (2025). Sovereign AI infrastructure - MMIV Conference 2025. Full text in Research Archive Show summary

  Infrastructure and competence building in AI/ML in Norway to power the future of the health sector

• Samarakoon, Pubudu Saneth & Razick, Sabry (2024). Accelerating Genomics Workflows on GPUs: Hands-on with NVIDIA Parabricks. doi: https:/github.com/OBIWOW/OBiWoW-2024/tree/main/12-Thursday/accelerating-genomics-workflows-on-gpus-hands-on-with-nvidia-parabricks. Full text in Research Archive Show summary

  In the fast-evolving world of genomics, Next-Generation Sequencing (NGS) analysis pipelines are pivotal for the comprehensive evaluation of patient genetic makeup in clinical and research settings. Traditional bioinformatic tools optimized for CPU-based computing often result in prolonged processing times that can extend to several days, creating significant bottlenecks as population genomics studies and clinical practices scale. To address these challenges, accelerated NGS pipelines leveraging graphical processing units (GPUs) have emerged as transformative solutions. This two-hour workshop aims to demonstrate the capabilities of NVIDIA Parabricks tools in NGS analysis. The session will commence with a concise overview of the NVIDIA Parabricks software suite, a leading toolset known for its remarkable ability to drastically reduce computational time while maintaining exceptional precision and recall in variant calling. Attendees will gain insights into the workflow of the Parabricks germline variant calling pipeline. Following the overview, participants will engage in a hands-on session designed to provide practical experience with Parabricks. The workshop will conclude with a brief discussion addressing the advantages and potential limitations of Parabricks. This workshop is ideal for bioinformaticians seeking to enhance their expertise and efficiency in NGS analysis through cutting-edge technological advancements.

• Razick, Sabry; Grimsmo, Andreas & Goodwin, Morten (2024). Computational resources. Full text in Research Archive Show summary

  Introduction to using high performance computing resources for Master level students

• Samarakoon, Pubudu Saneth; Ogul, Burcin Buket & Razick, Sabry (2024). Workshop: AI Tools with Biomedical Sciences (Norwegian AI Cloud). Full text in Research Archive
• Sugathan, Sherin & Razick, Sabry (2024). HPC Visualization using ParaView. Full text in Research Archive
• Razick, Sabry; Agueny, Hicham & Bjørgve, Magnar (2024). Tutorial: Accelerated Deep Learning via High-performance Computing (Norwegian AI Cloud). Full text in Research Archive
• Razick, Sabry & Ogul, Burcin Buket (2024). ParaView Workshop. Full text in Research Archive
• Razick, Sabry & Sætrom, Pål (2014). Data advertising and managing platform for biobanks - a use case for the eGenVar data management system. Full text in Research Archive
• Aranda, B; Blankenburg, H; Kerrien, S; Brinkman, FSL; Ceol, A & Chautard, E [Show all 40 contributors for this article] (2011). PSICQUIC and PSISCORE: accessing and scoring molecular interactions. Nature Methods. ISSN 1548-7091. 8(7), p. 528–529. doi: 10.1038/nmeth.1637. Full text in Research Archive
• Turinsky, Andrei L.; Razick, Sabry; Turner, Brian; Donaldson, Ian Matthew & Wodak, Shoshana J. (2011). Interaction databases on the same page. Nature Biotechnology. ISSN 1087-0156. 29(5), p. 391–393. doi: 10.1038/nbt.1867. Full text in Research Archive
• Razick, Sabry & Donaldson, Ian Matthew (2010). iRefIndex : Interaction Reference Index : International Conference on Intelligent Systems for Molecular Biology (ISMB), Stockholm, Sweden, June 27-July 02, 2009 : Poster E29. Universitetet i Oslo. Full text in Research Archive Show summary

  iRefIndex provides an index of protein interactions available in a number of primary interaction databases. This index allows the user to search for a protein and retrieve a non-redundant list of interactors for that protein. This system uses the Redundant Object Group (ROG) and Redundant Interaction Group (RIG) identifier to group proteins and interactions into redundant groups. This method allows users to integrate their own data with the iRefIndex in a way that ensures proteins with the exact same sequence and NCBI taxonomy identifier will be represented only once. iRefIndex has consolidated 10 different interaction databases using this method and provides a number of methods to access these data. This includes downloading as PSI-MI compliant tab-delimited file and from the iRefScape plugin for (http://www.cytoscape.org/), PSICQUIC web services and the iRefWeb(http://wodaklab.org/iRefWeb/). More details about iRefIndex and links available at: http://irefindex.uio.no/wiki/iRefIndex

• Razick, Sabry & Donaldson, Ian Matthew (2010). Design and prototype of a system to integrate and visualize biological interaction data : International Conference on Intelligent Systems for Molecular Biology (ISMB), Vienna, Austria: (July 21-25, 2007) : Poster. Universitetet i Oslo. Full text in Research Archive Show summary

  Biomolecular interaction data is an increasingly important bioinformatics dataset used to examine biological systems. However, these data are spread across multiple databases and expressed in disparate data structures and formats. A prerequisite for working with these data would be consolidation into a single non-redundant updated repository. An initial design and prototype for consolidating and visualizing interaction data will be presented in this poster with an especial emphasis on providing scalable and reliable web-services as part of the solution. The central point of operation is a data warehouse with numerous parsers retrieving updated information from existing data sources. We have designed parsers for PSI-MI 1.0, PSI-MI 2.5 XML files and tab delimited text files. The parsers for XML files are built using an event-driven pull-parsing API which gives them the ability to handle very large files. A local application interface provides access to this data warehouse for local programmers, java servlets and web services. The usage of various operating systems and programming languages by the intended clients were considered when designing the system. Therefore, platform independent protocols were used. Moreover, multiple implementations of hosting the services were considered with respect to their ability to handle large data sets reliably in a stateful manner. Using these web-services, a tool was developed to retrieve and present interaction data visually. The approach taken was to construct modules as plugins to existing molecular visualization software.

View all works in NVA