Comparison of the performance of multiple whole-genome sequence-based tools for the identification of Bacillus cereus sensu stricto biovar Thuringiensis

The Bacillus cereus sensu stricto (s.s.) species comprises strains of biovar Thuringiensis (Bt) known for their bioinsecticidal activity, as well as strains with foodborne pathogenic potential. Bt strains are identified (i) based on the production of insecticidal crystal proteins also known as Bt toxins or (ii) based on the presence of cry, cyt, and vip genes, which encode Bt toxins. Multiple bioinformatics tools have been developed for the detection of crystal protein-encoding genes based on whole-genome sequencing (WGS) data. However, the performance of these tools is yet to be evaluated using phenotypic data. Thus, the goal of this study was to assess the performance of four bioinformatics tools for the detection of crystal protein-encoding genes. The accuracy of sequence-based identification of Bt was determined in reference to phenotypic microscope-based screening for production of crystal proteins. A total of 58 diverse B. cereus s.l. strains isolated from clinical, food, environmental, and commercial biopesticide products were underwent WGS. Isolates were examined for crystal protein production using phase contrast microscopy. Crystal protein-encoding genes were detected using BtToxin_Digger, BTyper3, IDOPS, and Cry_processor. Out of 58 isolates, the phenotypic production of crystal proteins was confirmed for 18 isolates. Specificity and sensitivity of Bt identification based on sequences were 0.85 and 0.94 for BtToxin_Digger, 0.97 and 0.89 for BTyper3, 0.95 and 0.94 for IDOPS, and 0.88 and 1.00 for Cry_processor, respectively. Cry_processor predicted crystal protein production with highest specificity, and BtToxin_Digger and IDOPS predicted crystal protein production with the highest sensitivity. Three out of four tested bioinformatic tools performed well overall, with IDOPS achieving both high sensitivity and specificity (>0.90). IMPORTANCEBacillus cereus s.s. biovar Thuringiensis (Bt) is used as an organic biopesticide. It is differentiated from the foodborne pathogen Bacillus cereus s.s. by the production of insecticidal crystal proteins. Thus, reliable genomic identification of biovar Thuringiensis is necessary to ensure food safety and facilitate risk assessment. This study assessed the accuracy of WGS-based identification of Bt compared to phenotypic microscopy-based screening for crystal protein production. Multiple bioinformatics tools were compared to assess their performance in predicting crystal protein production. Among them, IDOPS performed best overall at WGS- based Bt identification.