The genome-wide Ensembl comparative genomics pipeline has changed significantly over 2008, and is now based on the Enredo-Pecan-Ortheus pipeline (EPO). These are a set of three programs which feed into each other. The Enredo programme (28) takes a set of genomes and creates a segmentation graph across all the genomes to extract a set of colinear homologous segments. Unlike the algorithms Ensembl has used previously, Enredo handles lineage specific duplications (for example, a duplication on the primate lineage giving rise to two copies of a series of genes in primates compared to other mammals). These colinear segments are then handed onto Pecan, a consistency based multiple aligner, which provides a highly accurate alignment of the homologous regions. Using an assessment based on ancestral repeats, Enredo+Pecan outperforms other combinations of alignment programs in mammals. Finally, the ancestral sequence reconstruction programme, Ortheus (29), generates accurate ancestral sequences across each region. Ortheus uses a branch transducer model, a type of HMM, to call deletion and insertion events, providing a realistic model under which it can infer the ancestral sequence. Figure 2 shows the
