The Structural Code of Breast Cancer Proteoform: Alternative Splicing-driven Protein Isoform Variation and Functional Diversification

Alternative splicing (AS) is widespread in cancer, yet current studies provide limited coverage of AS-derived proteoforms and lack a systematic, high-resolution atlas linking isoform sequence variation to structural remodeling and functional diversification across tumors. Here, we introduce the three-dimensional Structure Isoform Galaxy (3DisoGalaxy) platform, an isoform-resolved breast cancer knowledge base that integrates the transcriptome, translatome, and foldome to enable structure-grounded, large-scale computational analysis of AS-derived proteoforms. We integrated PacBio long-read RNA sequencing (Iso-Seq; n = 35), short-read RNA-seq from four breast cancer cohorts, and two ribosome profiling (Ribo-seq) datasets (n = 42) to curate full-length transcript variants and define translationally supported open reading frames (ORFs). Across these datasets, we identified 123,395 transcript variants and 73,715 ORFs through stringent transcript- and translational-level curation, enabling construction of a breast cancer foldome of high-quality protein structure models. A stringent, quality-controlled subset of 46,601 structures was further annotated with structure-resolved motifs and organized into a structural similarity network, enabling structure-based functional concordance analyses and forming 3DisoGalaxy. Graphical Abstract O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=139 SRC="FIGDIR/small/722115v1_ufig1.gif" ALT="Figure 1"> View larger version (51K): org.highwire.dtl.DTLVardef@ceb7d8org.highwire.dtl.DTLVardef@18802d7org.highwire.dtl.DTLVardef@72b1a3org.highwire.dtl.DTLVardef@1ec7bf4_HPS_FORMAT_FIGEXP M_FIG C_FIG 3DisoGalaxy generates testable mechanistic hypotheses, exemplified by a KRAS isoform, KRAS4A, showing selective loss of motif instances relative to the canonical isoform, and by nomination of a non-canonical AKT1 isoform with the strongest triple-negative breast cancer (TNBC)-biased expression shift within the AKT1 family and a modest relapse-free survival association. 3DisoGalaxy is accessible through an interactive web portal that provides integrated multi-omics results and 3D structure visualization.