timsRust: A Modular Foundation for Performant timsTOF Data Workflows

Significance of the topic

Trapped ion mobility spectrometry (TIMS) coupled to time-of-flight (TOF) mass spectrometry expands proteomics by adding an ion-mobility separation axis, increasing information content and enabling novel acquisition modes. However, TIMS-enabled workflows produce larger, more complex raw datasets that challenge existing processing pipelines. Efficient, transparent and extensible raw-data access and preprocessing are therefore critical to realize the full potential of timsTOF instruments in high-throughput proteomics and to integrate ion-mobility data into peptide-centric analysis tools.

Objectives and overview of the study

The work introduces timsRust, an open-source Rust-based framework designed to provide fast, modular, and memory-safe direct access to timsTOF raw data and to offer preprocessing primitives that simplify downstream peptide identification and quantification. Key objectives were to (1) eliminate conversion bottlenecks to mzML/MGF, (2) enable efficient centroiding and pseudo-MS2 generation for DIA data (diaPASEF), and (3) present a façade API for easy integration with search engines such as Sage and with language bindings (Python/C).

Methods and processing workflow

timsRust is implemented as a collection of modular Rust crates that separate core data types and traits from format-specific readers and preprocessors. The core design principles and processing steps are:

• Modular crate layout: shared primitives in a core crate, format-specific crates (e.g., tdf/tsf readers), optional SDK/external crates, and a façade crate that unifies implementations behind a single API.
• Memory-safe, concurrent processing: Rust's typing and ownership model used to enable safe multithreaded access and scale with available CPU cores.
• Frame-by-frame centroiding: each TIMS frame is reduced to centroids to lower data volume and noise while retaining peptide-relevant signals.
• Pseudo-MS2 generation for DIA: centroided fragments are linked to centroided precursors when ion mobility and isolation windows match, producing spectra compatible with DDA-style search engines.
• Interoperability: optional bindings (Python/C) and output compatible with peptide-centric tools such as the open-source Sage search engine.

Used instrumentation

The software is developed for and benchmarked on Bruker timsTOF raw data. Reported performance benchmarks used a laptop with 28 virtual cores. Integration partners include the Sage open-source search engine; timsRust supports format-specific readers for Bruker file types and provides centroiding and spectrum-generation crates.

Main results and discussion

Major findings and performance observations:

• Direct raw-data access: timsRust reads Bruker raw files directly during peptide-spectrum matching, eliminating the need for intermediate mzML or MGF conversion. Bypassing conversion removes a dominant I/O and CPU bottleneck.
• Speed improvements: end-to-end workflows using timsRust were approximately twice as fast as traditional conversion-based pipelines. Reading spectra from raw files was often as fast or faster than from converted formats.
• High-throughput centroiding and pseudo-MS2 generation: generating ~150,000 spectra from a 7-minute HeLa diaPASEF run required ~15 seconds on a 28-vCore laptop (~100 μs per spectrum); the complete Sage search workflow completed in ~30 seconds.
• Data reduction with retained interpretability: frame-by-frame centroiding significantly reduces noise and data volume while preserving peptide-identifying information, enabling conventional DIA tools or pseudo-DDA workflows to perform effectively, especially on short gradients.
• Extensibility and safety: the Rust implementation combines high performance with compile-time safety guarantees that reduce concurrency and memory errors in large-cohort analyses. Wrappers allow reuse from Python and C environments.

These results demonstrate that architectural choices—direct raw access, lightweight centroiding, and modular design—yield practical throughput gains while remaining compatible with established peptide-centric search engines.

Benefits and practical applications

Practical advantages of timsRust include:

• Substantial reduction of preprocessing overhead by avoiding intermediate file conversion.
• Faster end-to-end DDA and DIA pipelines on commodity hardware, enabling more rapid iteration and higher throughput in discovery and quantification experiments.
• Modular building blocks that facilitate integration into existing pipelines, custom preprocessing steps, or language ecosystems via bindings.
• Preservation of ion-mobility information in formats consumable by tools that do or do not natively support mobility, easing adoption across diverse software.

Future trends and opportunities

Potential directions and broader opportunities include:

• Wider community adoption and contributions to extend format support, preprocessing algorithms (e.g., adaptive centroiding, denoising), and connectors to other search engines and quantification tools.
• Standardization efforts for ion-mobility-aware data representations to improve interoperability across tools and repositories.
• Scaling to cloud and distributed environments for large-cohort studies, leveraging Rust's safety and performance for server-side deployments.
• Exploration of hardware acceleration (SIMD, GPU) for compute-bound preprocessing steps and real-time acquisition pipelines.
• Further development of ion-mobility-aware scoring and FDR control methods to fully exploit mobility in peptide identification and localization tasks.

Conclusion

timsRust provides a performant, memory-safe, and modular foundation for timsTOF raw-data access and preprocessing. By removing conversion bottlenecks, offering fast centroiding and pseudo-MS2 generation, and exposing a unified API with language bindings, timsRust accelerates both DDA and DIA proteomics workflows on standard hardware and simplifies integration with peptide-centric tools such as Sage. The framework’s design supports scalable, extensible analyses and paves the way for broader adoption of ion-mobility-resolved proteomics workflows.

References

1. Lazear M. Sage: An Open-Source Tool for Fast Proteomics Searching and Quantification at Scale. Journal of Proteome Research. 2023;22(11).