The complex mass spectra of entire proteins present significant challenges for data analysis in top-down proteomics (TDP). A key step in TDP analysis is spectral deconvolution, which extracts ion masses from spectral signals. Inaccuracies in this step can propagate to downstream analysis, such as proteoform identification and quantification, undermining the accuracy of the entire workflow. Despite its critical role, few robust methods have been introduced to estimate the false-discovery rate (FDR) for spectral deconvolution. We have developed a novel FDR estimation method specifically designed for spectral deconvolution. Our approach introduces decoy masses (DMs), artificially generated masses that mimic the behaviour of false positives (FPs) arising during the deconvolution process, which in turn enables the estimation of the FDR. We have validated our method using both in silico and experimental spectra. In silico datasets, our method achieved an average maximum absolute difference of 0.337 between estimated and true FDR across six datasets with varying resolution and noise levels. When applied to experimental spectra of known proteins, FDRs of approximately 5.64% and 9.75% were observed when 5% and 10% FDR thresholds were applied, respectively. The FDR estimation feature is implemented in the FLASHDeconv algorithm as part of the OpenMS open-source project.
