Artifact of the paper "What Do Users Ask in Open-Source Scientific Software Modules? A Study on GitHub Issues and Pull Requests using nf-core"

This study presents the first large-scale empirical investigation of user and developer interactions within a modular, automation-driven scientific software ecosystem. Focusing on the nf-core/modules repository, a cornerstone of the Nextflow-based bioinformatics community, the study analyzes 7,558 GitHub issues and pull requests collected between July 2019 and September 2025 to understand how contributors discuss, maintain, and evolve modular scientific workflows. Using a mixed-methods approach that integrates BERTopic modeling, process metrics, and statistical analysis, the study identifies 12 key discussion topics encompassing testing and environment setup, module refinement, planning and tracking new module additions, migration and expansion of modules, troubleshooting, and contributor onboarding. The topic taxonomy reveals that user discussions are shaped by nf-core’s strong governance model, automation infrastructure, and collaborative review culture. Quantitative analyses show that labeling and assignment practices significantly increase the likelihood of issue and PR closure, highlighting the importance of structured project management in sustaining open-source ecosystems. Moreover, migration and expansion tasks emerge as the most challenging activities, exhibiting the longest median resolution times and highest unresolved rates, reflecting the technical complexity and coordination effort required for template and dependency updates. The findings provide actionable insights for improving scientific workflow sustainability, emphasizing the need for enhanced automation, contributor guidance, and documentation clarity. Beyond nf-core, this work contributes to a broader understanding of human–automation collaboration in open-source scientific software and offers a reusable methodological framework for studying maintenance and community dynamics in similar ecosystems. Below, we provide the content for the replication. Dataset: We collect 9,145 GitHub issues and pull requests from the nf-core/modules repository (July 2019 – September 2025) using the data-collection.ipynb script. The data are shared under dataset/ModulesIssuesandPRs folder Preprocessing Scripts: data_preprocessing.ipynb scripts for text normalization, token cleaning, lemmatization (using spaCy), stopword removal (extended NLTK list), and removal of code snippets and URLs. Then linked_prs_removal.ipynb identifies Pull Requests directly linked with issues and finally produces a clean corpus suitable for topic modeling. Topic Modeling: topic_modeling_parameters_selection.ipynb determines the optimal parameters for the topic modeling and TopicModeling.ipynb generates the topic. The generated topics are shared under dataset/RQ1/Topic Distribution Statistical Analysis and Features Relationships: RQ2.ipynb and RQ3.ipynb contain the details for statistical and features analysis. Difficulty and Resolution Analysis: Aggregated topic-wise statistics including unresolved percentages and median resolution times (in hours) derived from GitHub metadata, supporting RQ4 analysis is shown in RQ4.ipynb The package supports full reproducibility of results reported in the paper, including data preprocessing, topic modeling, and statistical testing.Researchers can reuse the dataset and pipeline to study other open-source scientific software ecosystems or extend the topic-based taxonomy of nf-core discussions.