We study a method of mapping both mass spectra and sequences to feature vectors and the correlation between them. The method of calculating the feature vector from mass spectra is presented, together with a method for representing sequences. A correlation metric comparing both representations is studied. It shows strong correlation between two representation for the same peptides. It also demostrates that the effect of correlation is increased by using the longer sequences induced from the theoretical mass spectra. The method provides a promising step towards de novo sequencing. © Springer-Verlag Berlin Heidelberg 2005.