We can't find the internet
Attempting to reconnect
Something went wrong!
Hang in there while we get back on track
PEZy-miner: An artificial intelligence driven approach for the discovery of plastic-degrading enzyme candidates
AI summary Read the abstract
Researchers created PEZy-Miner, a machine learning framework using protein language models to identify plastic-degrading enzyme candidates across eleven plastic types, successfully concentrating known enzyme hits 14–30 times over random screening and identifying 27 new high-confidence candidates from 100,000 protein sequences.
Plastic waste has caused a global environmental crisis. Biocatalytic depolymerization mediated by enzymes has emerged as an efficient and sustainable alternative for plastic treatment and recycling. However, it is challenging and time-consuming to discover novel plastic-degrading enzymes using conventional cultivation-based or omics methods. There is a growing interest in developing effective computational methods to identify new enzymes with desirable plastic degradation functionalities by exploring the ever-increasing databases of protein sequences. In this study, we designed an innovative machine learning-based framework, named PEZy-Miner, to mine for enzymes with high potential in degrading plastics of interest. Two datasets integrating information from experimentally verified enzymes and homologs with unknown plastic-degrading activity were created respectively, covering eleven types of plastic substrates. Protein language models and binary classification models were developed to predict enzymatic degradation of plastics along with confidence and uncertainty estimation. PEZy-Miner exhibited high prediction accuracy and stability when validated on experimentally verified enzymes. Furthermore, by masking the experimentally verified enzymes and blending them into homolog dataset, PEZy-Miner effectively concentrated the experimentally verified entries by 14∼30 times while shortlisting promising plastic-degrading enzyme candidates. We applied PEZy-Miner to 0.1 million putative sequences, out of which 27 new sequences were identified with high confidence. This study provided a new computational tool for mining and recommending promising new plastic-degrading enzymes.
More Papers Like This
PlasticEnz: An integrated database and screening tool combining homology and machine learning to identify plastic-degrading enzymes in meta-omics datasets
AI summary Read the abstract
This study presents PlasticEnz, an open-source tool that combines Hidden Markov Models, DIAMOND alignments, and machine learning classifiers trained on ProtBERT embeddings to identify plastic-degrading enzymes in metagenomic datasets. Applied to plastic-exposed environments and field metagenomes, the tool successfully identified known PETases and distinguished plastic-contaminated from pristine sites.
Plastic-hydrolytic enzyme classification using explainable deep learning
AI summary Read the abstract
This research applies explainable deep learning methods to classify enzymes capable of breaking down plastic polymers through hydrolysis. By making the model's decision-making interpretable, the approach helps researchers understand which enzyme structural features drive plastic-degrading activity. The work advances the discovery and engineering of plastic-hydrolytic enzymes for biodegradation applications.
Machine Learning-Driven Multi-Objective Optimization of Enzyme Combinations for Plastic Degradation: An Ensemble Framework Integrating Sequence Features and Network Topology
AI summary Read the abstract
Researchers developed a machine learning framework to identify optimal enzyme combinations for breaking down polyester plastics. The study integrated kinetic data, protein sequence features, and network analysis to predict effective enzyme-substrate relationships, offering a computational approach to accelerating the discovery of enzymatic solutions for plastic waste degradation.
Widespread distribution of bacteria containing PETases with a functional motif across global oceans
AI summary Read the abstract
Using bioinformatics and AI-based protein structure prediction, researchers developed an algorithm to distinguish functional PET-degrading enzymes (PETases) from non-functional variants, identifying a high-confidence sequence motif for active PETases. Metagenomic analysis of 415 ocean samples revealed 23 PETase variants in nearly 80% of samples, predominantly between 1,000–2,000 meters depth and at the surface in high-pollution regions, with ocean PETases mainly encoded by Pseudomonadales bacteria.
Artificial intelligence-assisted mining of polyethylene terephthalate hydrolases
AI summary Read the abstract
Researchers reviewed how artificial intelligence is accelerating the discovery and engineering of PET-degrading enzymes, covering AI-driven high-throughput screening from biological databases, structure-function prediction, directed molecular evolution, and emerging de novo enzyme design — proposing an integrated framework to overcome the throughput bottlenecks of traditional enzyme mining approaches.
Research digests by email
When a large batch of papers lands in the Atlas, we read through it and send a short write-up of what stood out.