Predicting base editing outcomes with an attention-based deep learning algorithm trained on high-throughput target library screens

Kim F Marquart,Gerald Schwank,Michael Krauthammer,Anna Sintsova,Sharan Janjuha,Ahmed Allam,Nina Frey,Lukas Villiger

doi:10.1038/s41467-021-25375-z

Abstract

Base editors are chimeric ribonucleoprotein complexes consisting of a DNA-targeting CRISPR-Cas module and a single-stranded DNA deaminase. They enable transition of C•G into T•A base pairs and vice versa on genomic DNA. While base editors have great potential as genome editing tools for basic research and gene therapy, their application has been hampered by a broad variation in editing efficiencies on different genomic loci. Here we perform an extensive analysis of adenine- and cytosine base editors on a library of 28,294 lentivirally integrated genetic sequences and establish BE-DICT, an attention-based deep learning algorithm capable of predicting base editing outcomes with high accuracy. BE-DICT is a versatile tool that in principle can be trained on any novel base editor variant, facilitating the application of base editing for research and therapy.

Highlights

Base editors are chimeric ribonucleoprotein complexes consisting of a DNA-targeting CRISPR-Cas module and a single-stranded DNA deaminase
To capture base editing outcomes of SpCas[9] cytosine base editors (CBEs) and adenine base editors (ABEs) across thousands of sites in a single experiment, we generated a pooled lentiviral library of constructs encoding unique 20-nt sgRNA spacers paired with their corresponding target sequences (20-nt protospacer and a downstream NGG PAM site) (Fig. 1a)
Oligonucleotides containing the sgRNAs and corresponding target sequences were synthesized in a pool and cloned into a lentiviral backbone containing an upstream U6 promoter and a puromycin resistance cassette

Summary

Introduction

Base editors are chimeric ribonucleoprotein complexes consisting of a DNA-targeting CRISPR-Cas module and a single-stranded DNA deaminase. They enable transition of CG into TA base pairs and vice versa on genomic DNA. While base editors have great potential as genome editing tools for basic research and gene therapy, their application has been hampered by a broad variation in editing efficiencies on different genomic loci. A major limitation of base editors is their broad variation in editing efficiencies across different target sequences These can be influenced by several parameters, including the consensus sequence preference of the deaminase[4], and the binding efficiency of the sgRNA to the protospacer[5]. We develop a machine learning algorithm capable of predicting base editing outcomes of commonly used ABEs and CBEs on any given protospacer sequence in silico available via www.be-dict.org

Methods

Results

Conclusion