Machine learning applied to predicting microorganism growth temperatures and enzyme catalytic optima

Li, Gang; Rabe, Kersten S.; Nielsen, Jakob Skov; Engqvist, Martin K. M.

doi:10.1101/522342

Cited by 3 publications

(2 citation statements)

References 49 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The GTDB framework also enables linking phenotypes to genomes, as well as multiple sequences from the same genome across alignments, which can aid studies of protein and RNA complexes. We used a machine learning approach 38 to assign an optimal growth temperature to each reference genome in the GTDB, building on experimental measurements 36,37 . We then tested whether these temperatures, assigned to the RNAs identified in the GTDB genomes, could be used to identify thermophilic mutations that stabilize the E. coli ribosome.…”

Section: Discussionmentioning

confidence: 99%

“…However, since the TEMPURA and Gosha databases only include cultivated species, they only have experimental OGTs for 15% of the GTDB reference species. We therefore inferred OGTs of all GTDB reference genomes using TOME 38 . TOME predicts the OGT for an organism using a machine learning model trained on proteome-wide dipeptide (2-mer) distributions.…”

Section: Mapping Of Optimal Growth Temperatures To Gtdb Reference Gen...mentioning

confidence: 99%

See 1 more Smart Citation

RNA language models predict mutations that improve RNA function

Shulgina,

Trinidad,

Langeberg

et al. 2024

Preprint

View full text Add to dashboard Cite

Structured RNA lies at the heart of many central biological processes, from gene expression to catalysis. While advances in deep learning enable the prediction of accurate protein structural models, RNA structure prediction is not possible at present due to a lack of abundant high-quality reference data. Furthermore, available sequence data are generally not associated with organismal phenotypes that could inform RNA function. We created GARNET (Gtdb Acquired RNa with Environmental Temperatures), a new database for RNA structural and functional analysis anchored to the Genome Taxonomy Database (GTDB). GARNET links RNA sequences derived from GTDB genomes to experimental and predicted optimal growth temperatures of GTDB reference organisms. This enables construction of deep and diverse RNA sequence alignments to be used for machine learning. Using GARNET, we define the minimal requirements for a sequence- and structure-aware RNA generative model. We also develop a GPT-like language model for RNA in which triplet tokenization provides optimal encoding. Leveraging hyperthermophilic RNAs in GARNET and these RNA generative models, we identified mutations in ribosomal RNA that confer increased thermostability to the Escherichia coli ribosome. The GTDB-derived data and deep learning models presented here provide a foundation for understanding the connections between RNA sequence, structure, and function.

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Mapping Of Optimal Growth Temperatures To Gtdb Reference Gen...mentioning

confidence: 99%

RNA language models predict mutations that improve RNA function

Shulgina,

Trinidad,

Langeberg

et al. 2024

Preprint

View full text Add to dashboard Cite

show abstract

The era of big data: Genome-scale modelling meets machine learning

Antonakoudis

Barbosa

Kotidis

et al. 2020

Computational and Structural Biotechnology Journal

View full text Add to dashboard Cite

show abstract

Predicting protein domain temperature adaptation across the prokaryote-eukaryote divide

Johnson

Casstevens

et al. 2021

Preprint

View full text Add to dashboard Cite

Protein thermostability is important for fitness but difficult to measure across the proteome. Fortunately, protein thermostability is correlated with prokaryote optimal growth temperatures (OGTs), which can be predicted from genome features. Models that can predict temperature sensitivity across the prokaryote-eukaryote divide would help inform how eukaryotes adapt to elevated temperatures, such as those predicted by climate change models. In this study we test whether prediction models can cross the prokaryote-eukaryote divide to predict protein stability in both prokaryotes and eukaryotes. We compare models built using a) the whole proteome, b) Pfam domains, and c) individual amino acid residues. Proteome-wide models accurately predict prokaryote optimal growth temperatures (r2 up to 0.93), while site-specific models demonstrate that nearly half of the proteome is associated with optimal growth temperature in both Archaea and Bacteria. Comparisons with the small number of eukaryotes with temperature sensitivity data suggest that site-specific models are the most transferable across the prokaryote-eukaryote divide. Using the site-specific models, we evaluated temperature sensitivity for 323,850 amino acid residues in 2,088 Pfam domain clusters in Archaea and Bacteria species separately. 59.0% of tested residues are significantly associated with OGT in Archaea and 75.2% of tested residues are significantly associated with OGT in Bacteria species at a 5% false discovery rate. These models make it possible to identify which Pfam domains and amino acid residues are involved in temperature adaptation and facilitate future research questions about how species will fare in the face of increasing environmental temperatures.

show abstract

Machine learning applied to predicting microorganism growth temperatures and enzyme catalytic optima

Cited by 3 publications

References 49 publications

RNA language models predict mutations that improve RNA function

RNA language models predict mutations that improve RNA function

The era of big data: Genome-scale modelling meets machine learning

Predicting protein domain temperature adaptation across the prokaryote-eukaryote divide

Contact Info

Product

Resources

About