dnoise: Fast Native Data Reduction for Bruker timsTOF

The Bruker timsTOF system generates large .d files during data acquisition, which can be challenging to store, transfer, and archive. To address this issue, a new open-source Rust tool called dnoise has been developed. This tool efficiently removes points directly from timsTOF frames while retaining the crucial information needed for analysis.

By applying acquisition-aware gates, dnoise ensures that ions forming coherent streaks across the ion-mobility dimension are preserved. Additionally, it allows for the retention of signal that cannot be selected for fragmentation.

In a benchmark study involving three species, dnoise was tested on both ddaPASEF and diaPASEF modes at 5- and 15-minute gradients. The results showed that default MS1-only denoising reduced the frame binary size by 35 to 53% without compromising label-free quantification accuracy. Importantly, the peptide-spectrum-match, peptide, and protein-group counts remained unchanged, indicating that the MS/MS spectra were left untouched.

When it came to diaPASEF, only slight changes were observed in the precursor and protein-group counts.

The processing time for each run completed within 69 seconds on the benchmark workstation, even with the optional MS/MS denoising enabled. However, this optional denoising option did result in a slight decrease in identifications. Overall, the dnoise tool successfully removes a significant portion of the native timsTOF frame data while maintaining the analytical integrity of the samples.