Machine learning predicts new anti-CRISPR proteins

Abstract

The increasing use of CRISPR–Cas9 in medicine, agriculture, and synthetic biology has accelerated the drive to discover new CRISPR–Cas inhibitors as potential mechanisms of control for gene editing applications. Many anti-CRISPRs have been found that inhibit the CRISPR–Cas adaptive immune system. However, comparing all currently known anti-CRISPRs does not reveal a shared set of properties for facile bioinformatic identification of new anti-CRISPR families. Here, we describe AcRanker, a machine learning based method to aid direct identification of new potential anti-CRISPRs using only protein sequence information. Using a training set of known anti-CRISPRs, we built a model based on XGBoost ranking. We then applied AcRanker to predict candidate anti-CRISPRs from predicted prophage regions within self-targeting bacterial genomes and discovered two previously unknown anti-CRISPRs: AcrllA20 (ML1) and AcrIIA21 (ML8). We show that AcrIIA20 strongly inhibits Streptococcus iniae Cas9 (SinCas9) and weakly inhibits Streptococcus pyogenes Cas9 (SpyCas9). We also show that AcrIIA21 inhibits SpyCas9, Streptococcus aureus Cas9 (SauCas9) and SinCas9 with low potency. The addition of AcRanker to the anti-CRISPR discovery toolkit allows researchers to directly rank potential anti-CRISPR candidate genes for increased speed in testing and validation of new anti-CRISPRs. A web server implementation for AcRanker is available online at http://acranker.pythonanywhere.com/.

AcrDB: a database of anti-CRISPR operons in prokaryotes and viruses

“…The top 10% ranked proteins in the two runs were then intersected with the Acr-Aca operons from AcrFinder. Acr-Aca operons containing top 10% ranked proteins are considered to be more confident candidates according to the AcRanker paper ( 23 ). Meanwhile, after the AcrFinder step, all the 57 879 proteins (unique IDs) of the predicted Acr-Aca operons were processed by PaCRISPR to receive a prediction score (Figure 1 ) and those with a score >0.5 were considered as more confident candidates according to the PaCRISPR paper ( 24 ).…”

“…Additionally, in the year 2020, four standalone/webserver bioinformatics tools become available for automated Acr discovery (Table 1 ). Three of these tools were developed by authors of this paper: AcrFinder ( 22 ), AcRanker ( 23 ) and PaCRISPR ( 24 ). AcrFinder implemented a bioinformatics pipeline that was reported in ( 25 ), where we developed a bioinformatics pipeline to identify genomic operons containing Acr homologs and/or Aca homologs by combining three computational approaches: homology, GBA (guilt-by-association), and STS (self-targeting spacer) ( 26 ).…”

“…AcrFinder implemented a bioinformatics pipeline that was reported in ( 25 ), where we developed a bioinformatics pipeline to identify genomic operons containing Acr homologs and/or Aca homologs by combining three computational approaches: homology, GBA (guilt-by-association), and STS (self-targeting spacer) ( 26 ). AcRanker uses amino acid compositions of known Acr and non-Acr proteins to train an Extreme Gradient Boosting (XGBoost) ranking algorithm to rank putative Acr proteins in a given test proteome, and has been used to assist experimental discovery of novel Acr proteins ( 23 ). PaCRISPR builds position-specific scoring matrices (PSSMs) to capture evolutionary features in Acr alignments, and then trains a support vector machine (SVM) for Acr classification ( 24 ).…”

“…Briefly, combining AcrFinder ( 22 ), AcRanker ( 23 ), and PaCRISPR ( 24 ), we have performed a large-scale bioinformatics data mining for putative Acr-Aca operons in all sequenced genomes of bacteria, archaea, and their viruses. As a result, a new pre-computed Acr-Aca operon database AcrDB was developed.…”

Integration of prophages into CRISPR loci remodels viral immunity inStreptococcus pyogenes

“…In addition, some phages also produce anti-CRISPR (Acr) proteins that specifically inhibit Cas nucleases (Davidson et al, 2020). Although none have been isolated from S. pyogenes phages, multiple Acrs that prevent the function of different type II-A Cas9 nucleases with diverse mechanisms of inhibition have been reported (Eitzinger et al, 2020;Forsberg et al, 2019;Hynes et al, 2017;Mahendra et al, 2020;Rauch et al, 2017;Uribe et al, 2019). Acrs are typically present in temperate phages and are robustly transcribed in both the lytic and latent cycles immediately after infection (Bondy-Denomy et al, 2013;Stanley et al, 2019).…”