BRAKER | fully automated prediction of protein coding gene structures | Genomics library

BRAKER is a Perl library typically used in Artificial Intelligence, Genomics applications. BRAKER has no bugs, it has no vulnerabilities and it has low support. However BRAKER has a Non-SPDX License. You can download it from GitHub.

BRAKER mainly features semi-unsupervised, extrinsic evidence data (RNA-Seq and/or protein spliced alignment information) supported training of GeneMark-EX[F1] and subsequent training of AUGUSTUS with integration of extrinsic evidence in the final gene prediction step. However, there are now a number of additional pipelines included in BRAKER. In the following, we give an overview of possible input files and pipelines:. Figure 2: BRAKER pipeline A: training GeneMark-ES on genome data, only; ab initio gene prediction withAUGUSTUS. Figure 3: BRAKER pipeline B: training GeneMark-ET supported by RNA-Seq spliced alignment information, prediction with AUGUSTUS with that same spliced alignment information. Figure 4: BRAKER pipeline C: training GeneMark-EP+ on protein spliced alignment, start and stop information, prediction with AUGUSTUS with that same information, in addition chained CDSpart hints. Proteins used here can be of any evolutionary distance to the target organism. Figure 5: BRAKER pipeline D: training GeneMark-ETP+ supported by RNA-Seq alignment information and information from proteins (proteins can be of any evolutionary distance). Please be aware that GeneMark-ETP+ is still under development, BRAKER can currently execute a precursor of the mature version. Introns supported by both RNA-Seq and protein alignment information are treated as “true positive introns”, their prediction in gene structures by GeneMark-ETP+ and AUGUSTUS is enforced. Important: It is not always best to use all evidence! So far, we found this approach to work well for large genomes, but accuracy on small and medium sized genomes is unstable. Please have a look at the poster from PAG 2020 before running this pipeline. Figure 6: Additional pipeline B: training AUGUSTUS on the basis of spliced alignment information from proteins of a very closely related species against the target genome. Figure 7: Additional pipeline A: training GeneMark-ET supported by RNA-Seq spliced alignment information, prediction with AUGUSTUS with spliced alignment information from RNA-Seq data and with gene features determined by alignments from proteins of a very closely related species against the target genome. *Note:* This pipeline is deprecated since pipeline C can also use proteins of closely related species in addition to OrthoDB. Figure 8: Additional pipeline C: training GeneMark-ET on the basis of RNA-Seq spliced alignment information, training AUGUSTUS on a set of training gene structures compiled from RNA-Seq supported gene structures predicted by GeneMark-ET and spliced alignment of proteins of a very closely related species. *Note:* This pipeline is deprecated since pipeline D can also use proteins of closely related species in addition to OrthoDB.