To evaluate RATT, we assessed its performance using manually annotated genomes. The M. tuberculosis strain H37Rv (NCBI:AL123456) was used to annotate the genome of strain F11 using the ‘Strain’ comparison option. Results were compared with the existing annotation of F11 (NCBI:CP000717). In addition, the annotation of P. chabaudi was mapped to the P. berghei version 9 genome using the ‘Species’ comparison option. The P. chabaudi/P. berghei dataset can be downloaded from http://ratt.sourceforge.net/Chab_berg.zip . The files relating to the transfer of annotation between P. berghei assemblies can be found at http://ratt.sourceforge.net/Berg_berg.zip . The transfer was performed using the ‘Assembly.Repetitive’ parameters, and the results are included in the latest GeneDB version (http://www.genedb.org ).
Although, direct benchmarking was not possible because RATT presents a new strategy, we ran Glimmer3—a popular ab initio gene predictor—on the tuberculosis dataset as a comparator. Particular attention was given to the number of CDSs transferred or predicted, and whether their boundaries coincided with curated models. After running RATT on each of the three datasets, the transferred annotations were manually checked in Artemis and ACT.
Although, direct benchmarking was not possible because RATT presents a new strategy, we ran Glimmer3—a popular ab initio gene predictor—on the tuberculosis dataset as a comparator. Particular attention was given to the number of CDSs transferred or predicted, and whether their boundaries coincided with curated models. After running RATT on each of the three datasets, the transferred annotations were manually checked in Artemis and ACT.
