Key points are not available for this paper at this time.
As viral vectors consolidate their role as leading vehicles in therapeutic gene delivery, the demand for finely engineered variants with optimized properties continues to intensify. Computational approaches have emerged as powerful enablers of this effort, extending beyond the limits of rational design and large-scale mutagenesis to allow a deeper and more systematic exploration of the vast protein sequence space. Both machine learning- and non-machine-learning-based methods have been used to support viral vector bioengineering, targeting from stability refinement and structural modeling to predicting functional properties and generating novel variants. Here, we synthesize recent advances in computationally guided viral vector design, bridging algorithmic innovation with the experimental and practical realities of bioengineering. We discuss key factors that determine successful implementation of machine-guided protein design in this context and broader challenges and opportunities that will shape the field as it moves toward more integrated and holistic design strategies. This review highlights the transformative potential of computational design for viral vector bioengineering and invites researchers, even those with limited computational backgrounds, to join this new wave of innovation shaping the next generation of viral gene therapies. • Machine-guided approaches are revolutionizing protein design and transforming viral vector bioengineering. • Computational biology tools enable optimization through structural modeling and stability refinement. • Discriminative models predict functional properties, and generative models design new sequences from scratch. • Success in machine-guided design relies on integrating vector choice, design goals, target proteins, and data strategy. • Beyond proteins, computational design can extend to other vector components.
Rodrigues et al. (Sun,) studied this question.