Proteogenomics

Identifies previously unannotated protein-coding regions by directly mapping tandem mass spectrometry (MS/MS) spectra to custom protein databases derived from sample-specific genomic and transcriptomic data. This process reveals non-canonical translation events, including novel open reading frames (ORFs), alternative start sites, and exon-skipping products that are invisible to standard reference-based proteomics. The approach validates the translational evidence of somatic mutations and gene fusions, confirming that genomic aberrations are actually expressed as mutant proteins.

Uses empirical protein-level evidence to correct and improve the structural annotation of genomes. By searching spectral data against six-frame translated genomes, proteogenomics provides direct experimental validation for gene models, exon boundaries, and splice junctions. This capability is critical for re-annotating non-model organisms and clinical samples, where reference annotations are incomplete. It systematically identifies pseudogenes that are actually translated and corrects mis-annotated start codons, elevating the quality of the foundational genomic reference itself.

Directly links expression quantitative trait loci (eQTLs) and somatic copy-number alterations to their downstream protein abundance changes. By correlating genomic variants with quantitative proteomic measurements, this capability distinguishes between variants that are buffered at the protein level and those that cause significant functional dysregulation. It provides a mechanistic bridge from GWAS loci to actionable protein-level phenotypes, identifying the specific protein mediators of disease risk and revealing post-transcriptional regulation that uncouples mRNA levels from protein output.

Directly identifies MHC-bound peptides arising from somatic mutations, gene fusions, and non-canonical translation events using immunopeptidomics. This proteogenomic pipeline confirms that a predicted mutant peptide is actually processed, presented on the cell surface, and capable of eliciting an immune response. It dramatically improves the accuracy of personalized cancer vaccine design by filtering out computationally predicted neoantigens that lack direct proteomic evidence, focusing therapeutic development on the most clinically relevant targets.

Maps site-specific phosphorylation, acetylation, and ubiquitination events onto the exact protein domains and sequence contexts altered by genomic mutations. This capability reveals how a single amino acid substitution can create or destroy a kinase substrate motif, rewiring entire signaling networks. By integrating phosphoproteomics with whole-exome sequencing, it identifies mutant-driven signaling activation as a direct functional consequence of a genomic lesion, providing a systems-level view of oncogenic signaling that is not apparent from genomics alone.

Employs unsupervised methods like Multi-Omics Factor Analysis (MOFA) and Similarity Network Fusion (SNF) to decompose the high-dimensional variation across genomics, transcriptomics, and proteomics into a sparse set of latent factors. This reveals coordinated multi-omics signatures that define clinically meaningful patient subgroups and survival outcomes. The approach captures the principal sources of biological heterogeneity, separating technical noise from true molecular signal to build robust biomarker panels that are consistent across multiple molecular layers.

The core computational process of constructing a customized protein sequence database from genomic and transcriptomic data. Instead of using a generic reference proteome, this approach incorporates sample-specific variants—single amino acid variants (SAVPs), novel splice junctions, and RNA editing events—to identify peptides that would otherwise remain hidden. This is essential for discovering non-canonical open reading frames (ncORFs) and tumor-specific neoantigens.

The analytical goal of proteogenomics: detecting peptides that contain amino acid sequences differing from the reference genome. These arise from:

• Somatic mutations: Single nucleotide variants (SNVs) translated into mutant proteins.
• RNA splicing: Novel exon-exon junctions creating unique peptide sequences.
• Gene fusions: Chimeric proteins from structural rearrangements like BCR-ABL.
• RNA editing: A-to-I editing events leading to recoded protein sequences. Mass spectrometry data is searched against the customized database to confirm these variants are actually translated.

A critical feedback loop where mass spectrometry data provides translational evidence to correct and improve gene models. Peptides identified outside known coding sequences can validate predicted genes, correct exon boundaries, and discover entirely new protein-coding loci. This process, known as proteogenomic reannotation, is particularly impactful in non-model organisms and for refining the human proteome by confirming the translation of pseudogenes and long non-coding RNAs.

A direct clinical application of proteogenomics in cancer immunotherapy. By identifying patient-specific mutant peptides that are actually presented on the cell surface by Major Histocompatibility Complex (MHC) molecules, proteogenomics provides stronger evidence for neoantigen vaccine design than genomic prediction alone. Integrating immunopeptidomics—the direct mass spectrometry analysis of MHC-bound peptides—with exome sequencing confirms which predicted neoantigens are processed and presented, creating a highly actionable target list for personalized vaccines and adoptive cell therapies.

Proteogenomics uniquely links genomic variation to protein-level phenotypes. A cis-regulatory effect occurs when a genetic variant near a gene directly influences its own protein abundance. A trans-regulatory effect involves a variant in one gene controlling the protein levels of a distant gene. By correlating protein quantitative trait loci (pQTLs) with protein expression data, researchers can map the functional consequences of disease-associated genetic variants, distinguishing between transcriptional and post-transcriptional regulation mechanisms.

Proteogenomics is inherently a multi-omics discipline, requiring the fusion of genomics, transcriptomics, and proteomics from the same sample. Advanced integration methods like Multi-Omics Factor Analysis (MOFA) and Similarity Network Fusion (SNF) are used to find latent factors that explain variance across all layers. This systems-level view reveals that mRNA abundance often explains only ~40% of protein variance, highlighting the critical regulatory role of translation and protein degradation that can only be captured through integrated proteogenomic analysis.