Hart, D. O. | Search bioRxiv

bioRxiv2023

casmini-tool: a comprehensive database for efficient and specific guide RNA design using dCasMINI

The dCasMINI protein is a hypercompact, nuclease-inactivated CRISPR-Cas system engineered for transcriptional modulation and epigenetic editing [Xu et al., 2021]. The small size of dCas-MINI (529 amino acids), less than half the size of comparable Cas9 molecules, makes it ideal for AAV-based therapies which are frequently limited by AAVs small cargo capacity. Unlike Cas9 or Cas12a, there are no available computational tools for designing dCasMINI guides. To facilitate and accelerate the development of dCasMINI-based applications, we synthesized knowledge regarding dCasMINI guide design and built a website to assist researchers in designing optimal guides for dCasMINI-based experiments for transcriptional inhibition (CRISPRi) and activation (CRISPRa); to ensure that our tool would be useful for therapeutic guide design, in which a guides off-target safety profile is of paramount importance, we specifically optimized alignment parameters for high-sensitivity to comprehensively report genome-wide off-targets. To investigate dCasMINIs full protospacer adjacent motif (PAM) profile, we engineered libraries of PAMs and exhaustively characterized dCasMINIs ability to activate a locus with different PAMs. We also experimentally investigated the importance of each nucleotide position on the guide RNAs ability to activate its target, and characterized a 6bp high-fidelity seed region at the 5 end of the protospacer sequence which we identified to be intolerant to mismatches and deletions, and thus critical for true binding events. Taken together, our tool offers CRISPRi/a guide design for every protein-coding gene in the human genome along with comprehensive off-target prediction, incorporating the most up-to-date information about dCasMINIs full PAM and protospacer design rules. The tool is freely available to use at www.casmini-tool.com.

bioengineering↗

bioRxiv2023

Improving few-shot learning-based protein engineering with evolutionary sampling

Designing novel functional proteins remains a slow and expensive process due to a variety of protein engineering challenges; in particular, the number of protein variants that can be experimentally tested in a given assay pales in comparison to the vastness of the overall sequence space, resulting in low hit rates and expensive wet lab testing cycles. In this paper, we propose a few-shot learning approach to novel protein design that aims to accelerate the expensive wet lab testing cycle and is capable of leveraging a training dataset that is both small and skewed ({approx} 105 datapoints, < 1% positive hits). Our approach is composed of two parts: a semi-supervised transfer learning approach to generate a discrete fitness landscape for a desired protein function and a novel evolutionary Monte Carlo Markov Chain sampling algorithm to more efficiently explore the fitness landscape. We demonstrate the performance of our approach by experimentally screening predicted high fitness gene activators, resulting in a dramatically improved hit rate compared to existing methods. Our method can be easily adapted to other protein engineering and design problems, particularly where the cost associated with obtaining labeled data is significantly high. We have provided open source code for our method at https://github.com/SuperSecretBioTech/evolutionary_monte_carlo_search.

bioinformatics↗