Enhancement of protein thermostability by three consecutive mutations using loop-walking method and machine learning
Problem
Improving the heat stability of Burkholderia cepacia lipase without assaying all 8,000 possible triple mutants.
Why it matters
Thermostable enzymes are more useful in industrial processes that involve heat.
What the AI did
A machine-learning discrimination model trained on 214 mutants' thermostability data ranked 7,786 combinations in silico; 20 candidates were assayed.
Results
P233D/L234P/V235S mutant retained 66% activity after 60 °C for 30 min vs 5% wild-type and 35% prior mutant
Impact
ML-guided screening found a mutant with roughly 13-fold higher residual activity than wild-type after heat treatment.
Record
| Venue | Scientific Reports |
|---|---|
| Published | 2021 |
| Work type | article |
| Authors | Kazunori Yoshida, Shun Kawai, Masaya Fujitani, Satoshi Koikeda, Ryuji Kato, Tadashi Ema |
| Citations | 26 (OpenAlex, at import time) |
| References | 63 |
| Model or method | not stated in source |
| Organization | not stated in source |
| Primary source | https://doi.org/10.1038/s41598-021-91339-4 |
| Open access | gold |
| Keywords | thermostability, machine learning, lipase, mutagenesis, enzyme engineering |
Importance rating
real but narrower advance
Improves thermostability of one industrial enzyme via ML-guided screening; abstract demonstrates method on a single lipase with no stated scale of industrial use or impact.