This work presents a heuristic to the peptide sequencing in multiplex mass spectra. We developed a methodology that deconvolutes the multiplex mass spectrum, generating a modified specific spectrum for each peptide target mass. A directed spectrum graph is created from each deconvoluted spectrum and a cost function is defined to guide the depth first search algorithm through the sequence paths. The heuristic considers that the peptide with higher intensity would be correctly sequenced from the intact multiplex spectrum. A computational system called DNBuilder was developed. Mass spectra from a thyroid sample were used to test the methodology and compared with three well-known programs from the literature, as Peaks, pNovoPlus and Novor. The results show that our methodology enhances the correct amino acid sequencing amount in co-fragmented peptides spectra.