The cleanventory: Harmonized Global Chemical Inventory Data for Assessing Persistence and Mobility Hazards

Understanding the chemical diversity regulated across global trade markets is essential for informed decision-making and advancing regulatory frameworks. However, challenges such as limited public access to official documentation and diverse data formats hinder progress. With the growing number of regulated chemicals, a harmonized system aligned with the "one substance – one assessment" principle is urgently needed. Under the newly adopted CLP hazard classes for PMT (Persistent, Mobile, and Toxic) and vPvM (very Persistent and very Mobile), persistence and mobility are central to chemical risk assessment. Persistence under CLP refers to a chemical’s resistance to biodegradation, while mobility describes its potential to spread in the environment, particularly to water sources like groundwater and drinking water. These properties are crucial for regulatory frameworks, which aim to safeguard environmental and human health. The H2020 ZeroPM project (https://zeropm.eu) addresses these challenges through the development of the "cleanventory", an open-source global chemical inventory. Following FAIR principles (Findable, Accessible, Interoperable, Reproducible), the cleanventory features a user-friendly interface (https://database.zeropm.eu) and open-source code on GitHub. This database integrates harmonized data from literature, QSAR models, and regulatory sources, offering advanced analyses on persistence and mobility. It provides stakeholders with access to curated data and tools for independent replication. The cleanventory incorporates data from 38 files representing 25 inventories across 13 trade markets, encompassing over 990,000 entries, including 225,000 unique CAS Registry Numbers. Data cleaning and harmonization were critical for compatibility and quality. Structural information was inferred using CAS numbers and names via APIs (PubChem, CAS Common Chemistry, CCTE CompTox, and NCI/CADD), producing InChI strings ranked for probability using a weighted consensus approach. Persistence and mobility data were curated from scientific literature, QSAR models, and REACH dossiers and analysed using probabilistic Bayesian models to quantify uncertainties. The cleanventory includes over 344,000 unique InChI strings, with 126,000 identified as "most probable" structures. Initial assessments suggest that approximately 50% and 75% of entries exhibit persistence and mobility potential, respectively. Future expansions will include additional inventories and environmental exposure information. This robust database aligns with the EU Chemical Strategy for Sustainability, supporting regulatory processes with transparent, high-quality data.