DE NOVO Identification Of Novel Peptides From Large-scale Human Proteome Samples

Recently,many researches have reported that there are massive of functional peptides may be involved in disease development and progression.However,their identification in proteomics and characteristics stay unknown.If these peptides can be accurately identified in high throughput,it would sufficiently support the diagnosis and treatment.De novo sequencing is able to derive peptide sequence from protein mass spectrum directly,showing its superiority on novel peptide identification.However,its accuracy is lower than database searching for its limitation on spectrum quality and algorithm bias.In recent years,large-scale human proteome samples have been released by Clinical Proteomic Tumor Analysis Consortium(CPTAC),offering sufficient data and opportunity for researchers to establish and optimize strategies of de novo identification.Here a new strategy was proposed to identify novel peptides by the combination of de novo sequencing and re-database-searching on large-scale proteome samples.Based on the de novo sequencing result,102,855 reliable novel peptides and 952 mutated peptides were identified by logistic regression and re-database searching.Validation on three cancer cell lines(CAOV-3,HCT-116 and MCF-7)was conducted to prove their evidence in three omics(genomics,transcriptomics and proteomics).The result shows that 91.3% of novel peptides can be supported by at least one of omics.55.7% of mutated peptides can be validated in proteome of the cell lines,enriching the database of cancer protein mutations.Our research first proposes de novo identifying novel peptides using large-scale samples and make validation on independent cancer cell lines.The strategy is also instructive for novel peptide identification in other species.