On the Need for Mechanistic Models in Computational Genomics and Metagenomics

Abstract

Computational genomics is now generating very large volumes of data that have the potential to be used to address important questions in both basic biology and biomedicine. Addressing these important biological questions becomes possible when mechanistic models rooted in biochemistry and evolutionary/population genetic processes are developed, instead of fitting data to off-the-shelf statistical distributions that do not enable mechanistic inference. Three examples are presented, the first involving ecological processes inferred from metagenomic data, the second involving mechanisms of gene regulation rooted in protein–DNA interactions with consideration of DNA structure, and the third involving existing models for the retention of duplicate genes that enables prediction of evolutionary mechanisms. This description of mechanistic models is generalized toward future developments in computational genomics and the need for biological mechanisms and processes in biological models.

Introgression and repeated co-option facilitated the recurrent emergence of C4photosynthesis among close relatives

“…2017), and good mechanistic models exist for the evolution of DNA sequences (Liberles et al. 2013). Violation of model assumptions can still mislead the conclusions, but the multiplication of sources of information, coupled with the possibility to track the history of specific genes independently of the species tree, limits the risks of systematic errors.…”

“…However, the power of character modeling remains inherently limited by the small number of informative characters. Decomposing the phenotype into its components can solve this problem, especially when the underlying genetic determinism is considered (Oliver et al 2012;Niemiller et al 2013;Glover et al 2015;Meier et al 2017), and good mechanistic models exist for the evolution of DNA sequences (Liberles et al 2013). Violation of model assumptions can still mislead the conclusions, but the multiplication of sources of information, coupled with the possibility to track the history of specific genes independently of the species tree, limits the risks of systematic errors.…”

Spatial and Temporal Simulation of Human Evolution. Methods, Frameworks and Applications

“…Indeed, some of these simulators ignore recombination, which can bias evolutionary inferences [e.g., 118-121]. As a consequence, there is a need for more realistic computer simulators that implement complex substitution models of evolution, not only at the nucleotide level, but also at the codon [e.g., 43, 69], protein [e.g., 122-124] and genome-wide levels [88, 90]. Indeed, recombination (as well as other processes of exchange of genetic material) may generate evolutionary networks [67, 125] that should be considered to properly describe the history of human populations [see 18, 126].…”

Causes of evolutionary rate variation among protein sites

“…Phylogenetic models which allow for among-site rate heterogeneity universally provide better fits to data than do models which assume constant rates across sites 3;9-13 . However, such models are largely phenomenological in nature and contain no information about the mechanistic source of among-site rate heterogeneity 14 . Although it is clear that substantial rate variation exists, the underlying mechanisms which generate the observed rate heterogeneity remain elusive.…”