The study team presented BEguided, an integrated computational framework for variant-specific guide RNA design and for systematic comparison across multiple base-editor profiles. The platform bundles 38 predefined base-editor profiles but also accepts user-defined editors. BEguided integrates single-nucleotide variants directly into the target sequence rather than requiring customized reference genomes, which the authors reported simplifies pipeline setup for variant-centric design. The interface and underlying code are publicly available through the project website and linked GitHub repositories, according to the preprint.
BEguided emphasizes simultaneous consideration of several constraints that determine practical editing potential. PAM compatibility is assessed alongside editing-window positioning and the risk of bystander modifications. The investigators reported that balancing these factors is essential because each constraint can exclude otherwise plausible guide-target combinations. The platform therefore provides comparative outputs so users can prioritize editor-guide pairs that best satisfy user-defined stringency criteria.
The authors paired BEguided with a companion tool, BEguidedOFF, to address off-target risk in a base-editing–aware manner. BEguidedOFF performs off-target prediction that accounts for the specific base-editor context used in guide selection. The preprint indicates this tool is integrated into the workflow to enable users to filter candidate guides by predicted off-target profiles while retaining edit-context information—something standard off-target predictors may not incorporate. The source did not report experimental validation of BEguidedOFF predictions; the preprint describes the algorithmic approach and availability but does not provide bench-confirmation data within the abstract.
To illustrate the platform’s utility, the authors systematically assessed the accessibility of the 1,000 most frequent ABE-editable cancer mutations to adenine base editing. The analysis applied stringent filters for guide quality, bystander editing potential, and predicted off-targets. Under these strict criteria, only 7 mutations were deemed targetable when using SpCas9-ABE8e. The investigators then expanded the analysis to additional SpCas9-compatible adenine base editors; that broader set raised the count of potentially targetable mutations to 30.
The preprint reports stepwise gains in accessibility as PAM scope was relaxed and alternative Cas variants were considered. Including SaCas9 editors increased targetable mutations to 100. Factoring in relaxed-PAM variants such as Cas9-NG/SpG further expanded the list to 156 mutations. When the analysis allowed PAM-less targeting with SpRY, the number of potentially targetable mutations rose to 238. These results, as presented, quantify the substantial influence of PAM constraints and editor repertoire on the fraction of clinically relevant single-nucleotide variants that can be addressed computationally.
The cohort-level comparisons the authors reported make a clear operational point: choice of base editor can change whether a given single-nucleotide variant is practically targetable. Short editing windows, bystander bases within an editing window, and PAM restrictions all contributed to exclusion of many candidate mutations under the strict selection criteria used. The analysis therefore provides a prioritized map of which cancer-associated SNVs are reachable with current adenine base-editing tools, contingent on the filters applied for guide quality and predicted off-targets.
The authors made the data and scripts available in linked repositories, facilitating reproduction and further interrogation. The preprint’s figures and datasets—referenced in the article—offer researchers a resource to explore trade-offs between editor selection and mutation accessibility.
BEguided is accessible at the project website cited in the preprint, and code repositories for BEguided, BEguidedOFF, and associated datasets are listed as external links. The authors declared no competing interests in the manuscript.
The preprint does not report experimental bench validation of the computationally predicted editing opportunities. Nor does it supply clinical or in vivo validation data, detailed off-target assay results, or regulatory milestones. The work is presented as a computational prioritization framework; investigators reported counts of targetable mutations under defined stringency filters but did not describe downstream wet-lab confirmation in the abstract. The preprint itself has not been peer reviewed. Remaining open questions documented by the source include experimental corroboration of predicted edits, empirical off-target measurement for the suggested guides and editors, and translation of these computational priorities into validated therapeutic or diagnostic workflows.