When simulating from bacterial genomes, it appears that when no error profile is used (--model basic), it performs worse after classification on kraken2 than when a Miseq profile is used for instance.
--model basic is not an error-free error model. All bases have Q30 and all nucleotides have equal substitution rate. For error-free reads, you can use --model perfect.
When simulating from bacterial genomes, it appears that when no error profile is used (--model basic), it performs worse after classification on kraken2 than when a Miseq profile is used for instance.