Computational tools are widely used to boost the CRISPR-Cas design -- a brief introduction to the algorithms behind sgRNA design

Computational tools are widely used to boost the CRISPR-Cas design

    -- a brief introduction to the algorithms behind sgRNA design

The application of CRISPR-Cas as a genome editing tool utilizes the action of Cas9-sgRNA complex; the artificial single-guide RNA (sgRNA), a complex formed by the binding of the guide RNA(gRNA) to the trans-activating crRNA (tracrRNA), is designed to match and guide the Cas protein (usually Cas9) towards specific gene targets to be either inactivated or modified depending on the subsequent DNA repairing pathways. One consideration over the unwanted effects is about the off-target activities due to the partial paring between the sgRNA and random sequences resembling the anticipated one, as this may lead to potentially deleterious effects due to genetic alternations.

Multiple parameters exist in the experimental design of the CRISPR-Cas system, which comprise of the guide RNA design, the choice of Cas protein and the delivery method for the system etc. Currently, various stages of the design of CRISPR-Cas systems for research are assisted by corresponding computational approaches.

The major purpose of the computational tools is centered mainly on the selection of gRNA thus sgRNA. The initial stage of gRNA design can be the selection of useful gRNAs from the already existing libraries of gRNA online, e.g. CRISPRz for zebrafish, which provides information of gRNAs targeting the popular genetic sequences. The conduction of computational approaches allows for an efficient high-throughput large-scale screening of the performance of the gRNAs [1], thus greatly accelerate the designing process. For the customized design of gRNAs, the related software, for instance, CRISPOR, would search for the appropriate gRNA candidates and predict the quality of the selected gRNA from its efficacy of cleavage and specificity [1]. The latent relationships, between features of the gRNA sequence and its efficacy, have been suggested by various researches, while those relationships are utilized in the scoring methods for different computational models to infer the performance of the CRISPR-Cas system based on distinct gRNA candidates. To give an example, one of the properties of sgRNA synthesized, for instance the sequence GC contents, is implemented in both algorithms developed by Doench et al. and Morenno-Mateos et al. [1], where a linear regression model is trained to predict the activity of sgRNAs. The observed inconsistency among different predictive models is a result of various factors, including the difference in the experimental settings and activity assessment protocols applied in those models. As the off-target activities previously mentioned are expected to be eliminated, the current computational tools also incorporated algorithms that are able to evaluate the off-target activities of the Cas-sgRNA complex. There exist limitations at present, however, as most algorithms consider only sequence features of the srRNA, without covering critical factors involving chromatin contexts. Similarly, with inference to the sequence information of the gRNA and PAM, algorithms (SPROUT for example) can predict the occurrence of DNA repair mechanisms, being either nonhomologous end joining (NHEJ) or homology-directed repair (HDR), after the double stranded break is introduced by the Cas complex at the target site near PAM [1].

To summarize, there is still room for improvement for the present algorithms applied at each stage of CRISPR-Cas design, while this requires further insights into the influence of genetic factors like DNA structures and thermodynamics of nucleic acids against the action of Cas-sgRNA complexes [1,2]. And through a better design of the gRNA, it is expected that the off-target activities of the system can be reduced to achieve less adverse effects, thus making a better genome editing tool.

Reference:

 Aidan R O’Brien, Gaetan Burgio, Denis C Bauer, Domain-specific introduction to machine learning terminology, pitfalls and opportunities in CRISPR-based gene editing,Briefings in Bioinformatics, bbz145; 2020. https://doi.org/10.1093/bib/bbz145

Liu G, Zhang Y, Zhang T. Computational approaches for effective CRISPR guide RNA design and evaluation. Comput Struct Biotechnol J. 2019;18:35-44. Published 2019 Nov 29.; doi:10.1016/j.csbj.2019.11.006

Manghwar H, Li B, Ding X, et al. CRISPR/Cas Systems in Genome Editing: Methodologies and Tools for sgRNA Design, Off-Target Evaluation, and Strategies to Mitigate Off-Target Effects. Adv Sci (Weinh). 2020;7(6):1902312. Published 2020 Feb 6. doi:10.1002/advs.201902312

Sledzinski P., Nowaczyk M., Olejniczak M. Computational Tools and Resources Supporting CRISPR-Cas Experiments. Cells 2020, 9(5), 1288; 2020. https://doi.org/10.3390/cells9051288