Genome Annotation and Curation Using MAKER and MAKER-P

Abstract

This unit describes how to use the genome annotation and curation tools MAKER and MAKER-P to annotate protein-coding and noncoding RNA genes in newly assembled genomes, update/combine legacy annotations in light of new evidence, add quality metrics to annotations from other pipelines, and map existing annotations to a new assembly. MAKER and MAKER-P can rapidly annotate genomes of any size, and scale to match available computational resources.

Keywords: comparative genomics; gene finding; genome annotation; plants.

Figure 1. MAKER annotation workflow

MAKER masks repeats with RepeatMasker, aligns evidence to the genome with BLAST, polishes those alignments around splice sites using Exonerate, and runs a number of gene finders. MAKER also feeds evidence-based hints to the gene finders in order to improve their accuracy. These data are then synthesized into gene annotations.

Figure 2. Iterative gene finder training improves gene annotations

Using only EST data, and no gene-finders, MAKER annotates a single two-exon gene at this locus (MAKER_est2genome, purple). This annotation is consistent with the EST alignment (est2genome, beige), but is inconsistent with protein evidence data (protein2genome, yellow). After one round of SNAP training, MAKER annotates two models at this locus (MAKER_snap1, blue); these two models are more consistent with the protein evidence (protein2genome, yellow). An additional round of training yields a single MAKER annotation (MAKER_snap2, green) that is still more consistent the protein evidence. Note that SNAP is not improved on with further training (MAKER_snap3, orange).

Figure 3. Adding quality metrics to legacy annotations facilitates comparison between annotation sets

Shown on the y-axis is the cumulative distribution of AED for each dataset. The two legacy annotation sets are of comparable quality, with approximately 70% of their annotations having AEDs of less than 0.5 (orange and purple lines). Combining and updating legacy annotations results in a much improved annotation build (blue line), in which greater than 90% of the annotations have and AED less than 0.5.