Overview

These datasets accompany the publication Jewell et al. (2026). Each file contains the feature lists (in JSON format) from four different datasets in the NTSPortal database. This data is machine-generated and has not been validated, see also Section 5 of the data usage license and the additional data use disclaimer below. The field descriptions are included in the file "Document field descriptions for NTSPortal" (reproduced from the NTSPortal documentation).

The database has been updated since the publication (25.3 -> 26.1). For a detailed list of changes in each release of NTSPortal, please view the changelog.

Sampling and instrumental analysis

Excerpt from Jewell et al. (2026):

Sampling of water and SPM (suspended particulate matter) samples follows the methods described in Schlüsener et al. (2015) and Boulard et al. (2020), respectively. The analytical method for all samples is described in Nürenberg et al. (2015). A brief method description for both matrices follows. Water samples are filtered and injected without extraction, SPM samples are extracted with MeOH. Both are spiked with a mix of three deuterated internal standards (IS). The final concentrations are: bezafibrate-d4: 1 µg/L, olmesartan-d6: 0.5 µg/L, iopromide-d3: 2 µg/L [19]. From 2025 onwards, the Pharma-DX Mix 16 (Neochema GmbH, Bodenheim) is used, containing 16 IS including bezafibrate-d6, which is spiked to a final concentration of 1 µg/L. Analysis was performed on a LC-ESI-QqToF system (Agilent 1260 Infinity, Sciex TripleTOF 6600) equipped with a reversed phase, C18 column (Agilent Zorbax Eclipse Plus, 2.1 × 150 mm, 3.5 µm). Chromatographic gradient elution with water and acetonitrile (0.1vol% formic acid) was used.

Table 1: Sources of environmental samples

Data Processing

Excerpt from Jewell et al. (2026):

The data processing method was written in R, C++ and Python and is described in Jewell et al. (2020) and is available in the ntsportal repository (v25.3). Measurement files in proprietary vendor formats were converted to the open format mzXML using MSConvert v3.0.20301 (Kessner et al., 2008). The Collective Spectral Library (CSL 26.1) (Lessmann et al., 2026) was used to scan measurement files for matching m/z, retention times and MS² spectra. The results (initially stored as tables in R) were then reformatted and imported into Elasticsearch (a process known as ingesting) using the elasticsearch v8.18 Python package.

The library screening results (annotated features) were fetched from NTSPortal using the Elasticsearch API client in Python and written to JSON (one JSON file per compound). These were then bundled and compressed with Zip. 

Notes

The NTSPortal dashboard provides interactive search, analysis and data visualization of the datasets provided here. Access can be provided to universities, government agencies and non-profit organizations. Please send an account request to ntsportal@bafg.de.

Additional data use disclaimer

In the Licensor’s view, the disclaimer of warranty and liability pursuant to Section 5 of the license applicable here (Creative Commons Attribution 4.0 International) is invalid under German law. The Licensor therefore unilaterally and bindingly declares a waiver of the application of this Section 5 and will not invoke its validity against you/the users if located in Germany. Liability is therefore governed by statutory provisions. The providing entity is only liable in cases of intent or gross negligence (§ 521 BGB as well as §§ 839 BGB in conjunction with Art. 34 GG). No guarantee is assumed for the correctness, completeness, or timeliness of the data. Use is at your own risk.