Novel Proteoform Discovery by Precise Semi-De Novo Sequencing of Novel Junction Peptides

Alternative splicing allows a small number of human genes to encode large amounts of proteoforms that play essential roles in normal and disease physiology. Some low-abundance proteoforms may remain undiscovered due to limited detection and analysis capabilities. Peptides coencoded by novel exons and annotated exons separated by introns are called novel junction peptides, which are the key to identifying novel proteoforms. Traditional de novo sequencing does not take into account the specificity in the composition of the novel junction peptide and is therefore not as accurate. We first developed a novel de novo sequencing algorithm, CNovo, which outperformed the mainstream PEAKS and Novor in all six test sets. We then built on CNovo to develop a semi-de novo sequencing algorithm, SpliceNovo, specifically for identifying novel junction peptides. SpliceNovo identifies junction peptides with much higher accuracy than CNovo, CJunction, PEAKS, and Novor. Of course, it is also possible to replace the built-in CNovo in SpliceNovo with other more accurate de novo sequencing algorithms to further improve its performance. We also successfully identified and validated two novel proteoforms of the human EIF4G1 and ELAVL1 genes by SpliceNovo. Our results significantly improve the ability to discover novel proteoforms through de novo sequencing.