Genome editing technologies have extraordinary potential as new genomic medicines that address underlying genetic causes of human disease; however, it remains challenging to predict their long-term safety, because we do not know the consequences of potential side effects of genome editing such as off-target mutations or immunogenicity. Our long-term goal is to understand and predict such unintended biological effects to advance the development of safe and effective therapies. T-cells are an ideal cellular model because: 1) they are highly relevant as the most widely used cells for development of therapeutic genome editing strategies (such as cell- based treatments for HIV and cancer) and 2) mature T-cells encode a diverse T-cell receptor repertoire that can be exploited as built-in cellular barcodes for quantifying clonal expansion or depletion in response to specific treatments. We, therefore, propose the following specific aims: 1) to predict which unintended editing sites have biological effects on human T-cells by integrating large-scale genome-wide activity and epigenomic profiles with state-of-the-art deep learning models and 2) to develop a human primary T-cell platform to detect functional effects of genome editing by measuring clonal representation, off-target mutation frequencies, immunogenicity, or gene expression. If successful, our experimental and predictive framework will profoundly increase confidence in the safety of the next generation of promising genome editing therapies.
Genome editing technologies have extraordinary potential as the basis of new genomic medicines that address the underlying genetic causes of human disease; however, it is challenging to predict their long-term safety, because we do not know the consequences of potential unintended side effects of genome editing such as off- target mutations or immunogenicity. To define the biological effects of genome editing strategies, we will develop a human primary T-cell platform to sensitively detect functional effects coupled with an empirically-trained artificial intelligence models to predict them. Together, our platform will significantly improve confidence in safety assessments of promising genome editing therapeutics.
| Cheng, Yong; Tsai, Shengdar Q (2018) Illuminating the genome-wide activity of genome editors for safe and effective therapeutics. Genome Biol 19:226 |
