Abstract 2376 AI approaches for precision medicine of fusion oncogenes and oncoproteins

Fusion genes are the products of diverse types of structural variants. These chimeric genes have breakpoints inside the gene bodies by the chromosomal rearrangement initiated by the DNA double-strand breakage. The broken functional context of two genes provided unique tumorigenesis mechanisms and was used as the target of the tumor therapeutics. Throughout the long years of studies, we have tried to understand the mechanism and the functional effects of human fusion genes by developing many related tools. In this presentation, I will introduce multiple AI approaches we developed to advance the development of fusion gene-targeted therapeutics from diverse aspects of fusion genes and proteins. I.Fusion gene prediction – FusionAI: FusionAI leverages the known fusion breakpoints to provide a prediction model of the fusion genes from the primary genomic sequences via deep learning, helping researchers more accurately select fusion genes and better understand genomic breakage. II.Inferring the coding potential of fusion genes – deepORF: By retraining the model using a high-quality training data set, we could have dramatically increased the performance of the prediction of coding potentials. III.DNA damage repair system trace –Microhomology-mediated end joining (MMEJ), an error-prone DNA damage repair mechanism, frequently leads to chromosomal rearrangements due to its ability to engage in promiscuous end joining of genomic instability and also leads to increasing mutational load at the sequences flanking the breakpoints. We investigated the homology sequences around the genomic breakpoint area of human fusion genes, which were formed by the chromosomal rearrangements initiated by DNA double-strand breakage. IV.Functional annotation of fusion genes – FusionGDB2.0 has substantial updates of contents such as up-to-date human fusion genes, fusion gene breakage tendency score based on 20 kb DNA sequence around BP, investigation of overlapping between fusion breakpoints with 44 human genomic features, transcribed chimeric sequence and following open reading frame analysis with coding potential based on deep learning approach, and rigorous investigation of the protein feature retention of individual fusion partner genes in the protein level. V.3D structures and reliability check of fusion proteins – FusionPDB is organized into four levels: 43K fusion protein sequences (14.7K in-frame fusion genes, Level 1), over 2300 + 1267 fusion protein 3D structures (from 2300 recurrent and 266 manually curated in-frame fusion genes, Level 2), pLDDT score analysis for the 1267 fusion proteins from 266 manually curated fusion genes (Level 3), and virtual screening outcomes for 68 selected fusion proteins from 266 manually curated fusion genes (Level 4). VI.Fusion gene breakpoint-specific neoantigens – FusionNeoAntigen provides fusion gene-derived neoantigen prediction, virtual screening between fusion neoantigens and human leucocyte antigens (HLAs), fusion breakpoint RNA/protein sequences for developing vaccines, potential CAR-T targetable cell-surface fusion proteins. VII.Assessment of kinase in pan-cancer fusion genes – FusionNW: We suggest a novel way of assessing genes using a network propagation approach to infer how likely individual kinases influence the kinase fusion gene network composed of ∼ 5K kinase fusion gene pairs. To select a better seed of propagation, we chose the top genes via dimensionality reduction like a principal component or latent layer information of six features of individual genes in pan-cancer fusion genes. This work was partially supported by the National Institutes of Health grants R35GM138184 to P. Kim.