De novo sequencing is a spectrum analysis approach for mass spectrometry data to discover posttranslational modifications in proteins; however, such an approach is still in its infancy and is still not widely applied to proteomic practices due to its limited reliability. In this work, we describe a de novo sequencing approach for the discovery of protein modifications based on identification of the proteome UStags. The de novo information was obtained from Fourier-transform tandem mass spectrometry data for peptides and polypeptides from a yeast lysate, and the de novo sequences obtained were selected based on filter levels designed to provide a limited yet high quality subset of UStags. The DNA-predicted database protein sequences were then compared to the UStags, and the differences observed across or in the UStags (i.e., the UStags' prefix and suffix sequences and the UStags themselves) were used to infer possible sequence modifications. With this de novo-UStag approach, we uncovered some unexpected variances within several yeast protein sequences due to amino acid mutations and/or multiple modifications to the predicted protein sequences. To determine false discovery rates, two random (false) databases were independently used for sequence matching, and ∼3% false discovery rates were estimated for the de novo-UStag approach. The factors affecting the reliability (e.g., existence of de novo sequencing noise residues and redundant sequences) and the sensitivity of the approach were investigated and described. The combined de novo-UStag approach complements the UStag method previously reported by enabling the discovery of new protein modifications.The UStag method for unambiguous peptide and polypeptide identification has recently been demonstrated for the analysis of enzymatically (e.g., tryptic) digested cell lysates 1 and for the determination of natural intracellular proteolysis (degradation) of proteins 2 using accurate Fourier-transform tandem mass spectrometry (FT-MS/MS) data. Sequences are determined to be UStags when the accurately measured consecutive fragments reveal these sequences to be unique in the genome for single proteins. The UStags reported 1,2 are assigned for the candidates that have the top closest spectral similarities to the MS/MS measurement (e.g., candidates ranked from Sequest). Advantage of such a database search-UStag approach is that it produces sequence identities with extremely low false discovery rates for peptides/polypeptides having a large range of lengths and with various amino acid termini. 1,2 Also, this approach is capable of identifying unknown or unexpected changes, deviations, and errors from the predicted protein sequences. 1 However, the amino acid changes, deviations, and errors either on the UStag's prefix (i.e., the part of sequence prior to a UStag in the sequencing direction) or within In this work, we developed a de novo-UStag approach for the identification of protein posttranslational modifications (PTMs). On the basis of the assignments of UStags, th...