AcRanker

AcRanker predicts and ranks candidate anti-CRISPR proteins from protein sequence data using a machine learning ranking model.

Key Features:

• XGBoost-Based Ranking Model: Implements a gradient boosting model using XGBoost to prioritize candidate anti-CRISPR proteins from sequence data.
• Sequence-Based Prediction: Identifies anti-CRISPR candidates directly from protein sequence information without requiring additional genomic context.
• Proteome-Wide Candidate Ranking: Applies the trained model to proteomes to rank potential anti-CRISPR genes for experimental validation.
• Model Validation Framework: Evaluates predictive performance using non-redundant cross-validation and external validation datasets.
• Discovery of Anti-CRISPR Proteins: Enabled identification of previously unknown anti-CRISPR proteins including AcrIIA16 (ML1) and AcrIIA17 (ML8).

Scientific Applications:

• Anti-CRISPR Protein Discovery: Identifies novel inhibitors of CRISPR–Cas systems from genomic and proteomic datasets.
• CRISPR-Cas Regulation Studies: Supports investigation of proteins that inhibit Cas9 variants such as Streptococcus iniae Cas9 (SinCas9), Streptococcus pyogenes Cas9 (SpyCas9), and Staphylococcus aureus Cas9 (SauCas9).
• Genome Editing Control: Facilitates discovery of regulatory proteins for controlling CRISPR-Cas9 activity in biotechnology and synthetic biology.

Methodology:

AcRanker trains an XGBoost gradient boosting model on datasets of known anti-CRISPR proteins and applies the model to proteome sequences to rank candidate anti-CRISPR genes, with performance evaluated using non-redundant cross-validation and external validation.