phastCons

PhastCons is a computational tool within the PHAST (Phylogenetic Analysis with Space/Time) package designed to identify evolutionarily conserved elements in a multiple sequence alignment. It uses a phylogenetic hidden Markov model to distinguish conserved from non-conserved regions across related species, producing base-level conservation scores and predicted conserved elements.

The method requires a multiple sequence alignment of orthologous regions and a corresponding species tree, along with a substitution model. PhastCons employs a two-state continuous-time HMM, with a conserved state that has a lower substitution rate and a non-conserved (neutral) state with a higher rate. It accommodates rate variation across sites by discretizing a gamma distribution into several categories. The model yields posterior probabilities that each base belongs to the conserved state, which form the basis of the conservation scores.

The primary outputs are per-base phastCons scores, which represent the posterior probability of conservation, and a set of predicted conserved elements derived by applying thresholds to these scores and minimum length criteria. These outputs can be used directly in genome browsers and for downstream analyses in comparative genomics.

Applications of PhastCons focus on annotating conserved elements across genomes, including conserved noncoding elements, regulatory regions, and other functional elements, particularly in vertebrates and other lineages. The method was introduced by Siepel and colleagues in 2005 and remains a widely used approach for genome annotation and comparative genomics research.