Six New Research Projects Funded at USI

Six research projects carried out at the Università della Svizzera italiana (USI) have recently been awarded funding under the Open Research Data and Young Researchers calls, the latter promoted by the Institutional Research Fund (FIR). Below is a list of the six winning projects with a brief description.

Three projects have won under the Open Research Data call. Among them is the project “COVID-19 Data Hub”, led by Emanuele Guidotti, Postdoctoral Assistant at the Faculty of Economics, which received funding of 47,300 CHF. The project arose from the need to unify the vast amount of heterogeneous data released by governments around the world during the COVID-19 pandemic, the first pandemic of the digital age. The COVID-19 Data Hub aims to provide the research community with a unified dataset useful for better understanding the disease and its societal impact. This dataset has been updated hourly via automated pipelines since early 2020. However, governments around the world have recently changed their reporting criteria and introduced updates that cause the automatic pipelines to fail. Access to this data at scale risks becoming increasingly difficult without maintaining data aggregators that ensure high levels of interoperability and persistent archiving. This project aims to finalise a global, fine-grained epidemiological dataset. In particular, it has three goals: updating the COVID-19 Data Hub with the latest data by adapting to the changes in reporting criteria; validating the data through manual curation; and releasing the final version of the dataset. All integration and validation methods will be based on the design of the original data hub. The resulting dataset will be a unique resource for studying the first pandemic in the digital era. This work may also contribute to the development of best practices and standards for open research data, potentially increasing USI’s international visibility and promoting synergies with academic programmes.

The second funding, amounting to 42,000 CHF, was granted to the project “FMdb: the database to perform funnel-metadynamics calculations on GPCR systems”, conducted by Professor Vittorio Limongelli, Full Professor at the Faculty of Biomedical Sciences. Funnel Metadynamics (FM) is a powerful technique that elucidates the binding mode and mechanism of a ligand to its molecular target, while also precisely estimating the binding free energy. FM has been developed and enhanced by Professor Limongelli’s research group. It has been successfully applied to study numerous ligand/protein systems, but its broader use is hindered by the complexity of the simulation setup. This project aims to create a new open-access database of ready-to-use FM input files, called FMdb, focusing on ligand binding to G protein-coupled receptors (GPCRs). GPCRs are transmembrane proteins involved in conveying extracellular signals into the cellular environment. These receptors play a crucial role in various physiological and pathological processes and are highly relevant pharmaceutical targets. The functional mechanism of GPCRs can be controlled by orthosteric or allosteric ligands, which can trigger transitions of GPCRs into active or inactive states. GPCR complexes can be classified based on the ligand binding mode at extracellular, intra-membrane, and intracellular levels. This project will create a database of FM input files for binding free energy calculations at both orthosteric and allosteric sites for all experimentally resolved GPCR structures. The developed database will support the scientific community by providing ready-to-use FM inputs and will promote the adoption of FM in drug discovery studies.

Lastly, funding of 45,000 CHF was awarded to the project “REVITRANN – Workflows to reuse video data, transcripts and annotations in Interactional Linguistics”, led by Professor Johanna Miecznikowski-Fuenfschilling, Associate Professor at the Faculty of Communication, Culture and Society. Primary spoken language data (recordings) and secondary data at various levels of abstraction (transcripts, annotations) are now digitally encoded and can be shared in accordance with FAIR principles, provided that informed consent has been obtained from participants and appropriate data protection measures are in place. Transcribed video recordings have become increasingly important in the study of spoken language and in the social sciences. They capture key aspects of social interaction and spoken language as a multimodal resource. However, they are particularly complex to manage and demanding in terms of storage space. To make transcribed video data fully FAIR, including reusability, an infrastructure is needed to support online analysis in an integrated digital environment. Potential tasks a researcher might wish to carry out online include the use of personal or group workspaces; re-transcription or additional transcription (e.g. multimodal); comic book-style transcription; subtitling; time- and text-related annotation of varying complexity; storage and indexing of data segments; the creation of collections and memos; quantitative analysis; and download/export functions. This proposal contributes to building digital infrastructure for transcribed video data in Switzerland by reviewing existing platforms worldwide, describing the VideoScope software (LiRI/UZH) and its implementation at USI, and outlining the software’s development potential regarding analytical workflows for interactional linguistics.

The Institutional Research Fund (FIR) has also funded three projects under the Young Researchers call. The first funding, amounting to 97,800 CHF, was awarded to Andrea Rosà, Research Associate at the Faculty of Informatics at USI, for his project “Understanding and Mitigating Performance Variability on Managed Runtimes”. Performance variability is a well-known and harmful phenomenon, considered a long-standing, important, and open research problem both in academia and industry. It causes severe and unexpected performance degradation in multiple runs of the same application, without apparent reasons. Many programs are subject to variability even if they are single-threaded, have identical input across runs, and are executed in stable environments. Therefore, this phenomenon cannot be ignored. This project addresses the issue with the aim of understanding and mitigating variability, focusing particularly on applications running on managed runtimes. The project has three main goals: identifying major variability patterns that significantly impact application performance and the conditions under which they occur; determining the root causes of variability; and designing and implementing new strategies to mitigate variability. As a result, a framework will be designed and implemented for arbitrary applications running on managed runtimes, to automatically detect and explain their variability and suggest effective mitigation strategies.

The second project, funded with 120,000 CHF, is conducted by Matteo Pegoraro and is titled “Using Topological Data Analysis to Understand Microglia Shape Variability in Space and Time”. The aim of the project is to advance the state of the art in the morphological analysis of microglia through three approaches. Understanding microglial behaviour is essential for revealing the systems regulating brain evolution and adaptation mechanisms. This, in turn, has profound implications for diagnosing and treating neurodegenerative diseases such as Alzheimer’s and Parkinson’s, and age-related cognitive decline. To this end, researchers aim to improve the analysis of microglia from three perspectives: obtaining better descriptors of microglia morphology that convey more spatial structure information; enhancing exploration of microglia shape space using advanced tools from non-Euclidean statistics and topological machine learning; and analysing the dynamics of microglia adaptations over short time scales using optimal transport to describe cell adaptations.

Finally, funding of 199,700 CHF was awarded to the project “EverTest: Continual Learning for System-level Software Testing”, led by Matteo Biagiola, Research Associate at the Faculty of Informatics at USI. System-level testing is a critical activity to prevent severe issues from reaching end users. Manual writing of system-level tests is time-consuming and costly, and automated test generators are promising tools to reduce testing effort. However, current system-level test generators are not designed to handle evolving systems. Moreover, they generate static tests consisting of precise instruction sequences to interact with the system, which break easily when the system evolves and must be manually maintained. The EverTest project aims to address both limitations by proposing a test agent that evolves alongside the system under test. When the system evolves, the agent reuses the knowledge gained in previous test interactions to exercise newly introduced features while maintaining the ability to test unchanged ones. The test agent replaces the static system-level test suite, allowing developers to exercise specific behaviours during system evolution. The agent is easier to maintain and update than a static test suite, as its internal knowledge is automatically updated during training.