journals.asm.org
Psychrophiles elude genomic prediction of optimal growth temperature | mSystems
Most prokaryotic taxa remain difficult to grow and study in culture, and their environmental
preferences remain largely unquantified (1). As one example, we know that the optimal temperature for growth (OGT) can vary
dramatically across prokaryotes, but the OGT values for most prokaryotic species found
on Earth remain unknown. As OGT is an important determinant of when and where species
will grow, being able to reliably infer OGT values is useful for many reasons, from
identifying effective culturing strategies to predicting how the distributions of
microbial taxa will shift in response to climate change (1). Because the phylogenetic coherence of this trait is weak, with even members of
the same genus having variable OGTs (2), we cannot simply predict OGT values from taxonomic or phylogenetic information
alone (3). Instead, given the vast quantity of genomic information now available for uncultured
prokaryotes, there is broad interest in developing genome-based models to infer OGT
for uncultured species. Pre-existing modeling efforts have leveraged various genomic
features to infer OGT values with varying degrees of success. Examples of such models
include those based on nucleotide sequences (4–10); amino acid sequences and protein annotations (8, 11–14); and a combination of features derived from DNA and protein data (15). Yet, because trait axes orthogonal to OGT also impact genome composition, composition
alone is not sufficient to explain OGT differences between thermophiles and psychrophiles
(16, 17).
