Data Files for TIMS-Bench Towards community standards for benchmarking untargeted trapped ion mobility metabolomics tools and datasets

Title
TIMS-Bench datasets and benchmarking workflow outputs for untargeted trapped ion mobility metabolomics

Description / Abstract
This deposition contains the processed benchmarking datasets and reference files used in TIMS-Bench, a reproducible framework for evaluating untargeted TIMS metabolomics software across harmonization, annotation, and downstream performance metrics. The release includes harmonized feature tables, annotation outputs (cosine and spectral entropy), DreaMS-compatible embeddings, and curated ground-truth and library resources.

The dataset supports cross-tool benchmarking for MetaboScape, MS-DIAL, and MZmine, and is organized to enable direct reuse of the analysis notebooks and workflow components described in the associated manuscript (Rajkumar et al., 2026).

Process Summary (How the data were generated and organized)

1. Raw tool exports were harmonized into a common schema (feature metadata, spectral fields, and sample intensity columns).
2. Harmonized files were annotated using multiple similarity methods, including cosine and spectral entropy workflows.
3. Embedding-based resources were generated/collected for DreaMS-based analyses.
4. Ground-truth benchmark datasets were assembled for ReFRAME, NIST SRM, and plant spike-in evaluations.
5. Reference library files were prepared to support annotation and false-positive structural similarity analyses.

More details can be found in the GitHub repository: https://github.com/enveda/tims-bench

Data Contents

Top-level data groups:

• cache
• public_dataset
• groundtruth_dataset
• library_spectra

Public datasets included (10):

• MSV000084402
• MSV000090327
• MSV000091642
• MSV000095813
• MSV000096189
• MSV000096291
• MSV000097015
• MSV000097967
• MTBLS12332
• ST002402

Ground-truth datasets included (3):

• MSV000098263
• NIST_SRM
• plant_spikein

Reference/library files included:

• all_sorted_library_spectra.parquet (about 2.2 GB)
• nist_srm_spikein_lib.pq
• plant_spikein_lib.pq
• reframe_ms2s_with_ccs.parquet
• reframe_spikein_lib.pq
• reframe_smiles_list.csv

Common per-dataset structure:

• raw (original tool exports)
• harmonized (standardized parquet outputs)
• annotated_cosine_similarity
• annotated_spectral_entropy
• annotated_dreams_similarity (where available)
• embeddings

Intended Reuse
This deposition is intended for:

• Reproducible benchmarking of untargeted TIMS metabolomics tools
• Method comparison across annotation strategies
• Development and validation of new harmonizers and scoring methods
• False-positive and structural-similarity analyses against curated ground truth