Liftoff: A pipeline for mapping genes onto high-quality genome assemblies

Long–read sequencing and advanced computational methods have enabled rapid assembly of high quality genomes. Finding genes in these new assemblies is a necessary step to understating biologically important genetic variation within species. In organisms with an existing annotation, this is most easily done by lifting over the reference annotation onto the new assembly. Existing tools work by aligning full chromosomes to one another to find long syntenic regions, which only works well when the genomes are very similar. Also, these tools lift over each chromosomal interval individually which often results in exons mapping at locations that are inconsistent with the gene model, or homologous features mapping to the same locus. Liftoff overcomes these challenges by aligning gene sequences rather than full chromosomes and by considering the complete annotation when lifting over features. While many genes have a single complete alignment, others align in pieces or to many locations. When choosing the best mapping for these genes, Liftoff considers sequence identity, as well as the location of neighboring features and already-mapped homologs. Liftoff successfully mapped 99.6% of genes from the human genome version GRCh37 to GRCh38. To confirm that these mappings are correct, we checked the translation of the coding sequences and found that 98.6% of them match GRCh37 exactly.