D4.3 Code security and quality assessment

Nikolaos Nikolaidis, Dimosthenis Natsos

This document represents a key technical contribution to the RAISE project, specifically focusing on the quality and security of the distributed code in the infrastructure. To ensure the highest standard of scientific rigor and to facilitate the wide dissemination of project results, this deliverable has been prepared in the form of an academic research paper. The work described here is intended for submission to a peer-reviewed scientific journal in the field of Software Engineering. The paradigm shift towards Open Science has emphasized the need for Findable, Accessible, Interoperable, and Reusable (FAIR) data. However, for research to be truly reproducible, the software used to analyze this data must adhere to similar standards of quality. In the context of the RAISE project, which proposes a distributed crowdsourced data processing system where processing algorithms “travel” to the dataset, the quality of these portable scripts becomes critical to ensure both the long-term reproducibility of scientific results and the safety of the underlying infrastructure. This paper presents a methodology and a microservice-based tool designed to assess the maintainability and security of research software within the RAISE ecosystem. To validate this approach, we conducted an empirical study of 253 Python-based scripts across diverse scientific domains (e.g., Health, Transportation, Environment). Our results reveal that scientific scripting is structurally homogeneous across disciplines, indicating that maintainability challenges are universal rather than domain specific. However, security exposure is strongly correlated with the application domain, scripts in the Health/Medicine sector exhibit a disproportionately high frequency of critical vulnerabilities. Furthermore, computational methodology acts as the primary driver of infrastructure performance, with Machine Learning tasks showing significantly higher resource demands and execution variability. These findings underscore the necessity of context-aware quality gates in distributed scientific infrastructures to ensure the security, reproducibility, and sustainability of Open Science.