AbstractMost protein-coding genes in humans and other eukaryotes are made up of a collection of exonswhich are concatenated to form the messenger RNA (mRNA) that encodes a final proteinproduct. The well-known phenomenon of alternative splicing makes it possible for a single geneto encode multiple protein products by conditionally including only a subset of the genes exonsinto the expressed mRNA. A more surprising mechanism for producing alternate protein productsis to utilize an alternate reading frame of a standard exon through aberrant splicing; usingcustom software built in our research group we have found that this mechanism appears to bequite common. Specifically ~13% of all human genes include at least one exon that conditionallyencodes alternate peptides and these dual-coding exons are highly-conserved: 98%correspond to homologous exons in the mouse genome that also encode two open readingframes. Light exploration has identified dozens of human genes that show tissue-specificpatterns of reading frame usage suggesting a functional role for at least some of these variants.Here we describe a plan to (i) leverage massive public atlases of human tissue-specific anddevelopment-specific RNA-Seq and mass spectrometry data to tabulate the extent of differentialuse of these frame-shifted splicing variants and to (ii) analyze the computationally-predictedstructural and functional impact of dual-coding variants and the sequence signals controllingthem. The results of these analyses will be accumulated for release in an open and accessibleweb service.