publications
(* denotes equal contribution). All cover artworks are generated by DALL·E 3 with minimum cherry picking.
2026
-
Predicting genome-wide functional constraints with GPN-StarChengzhong Ye*, Gonzalo Benegas*, Carlos Albors, Jianan Canal Li, Sebastian Prillo, and 3 more authorsNature, 2026Genomic language models have emerged as a powerful approach for learning genome-wide functional constraints directly from DNA sequences. However, standard genomic language models adapted from natural language processing often require large model sizes and computational resources, yet still fall short of classical evolutionary models in predictive tasks. Here we introduce a genomic pretrained network with species tree and alignment representations (GPN-Star), which is a biologically grounded genomic language model featuring a phylogeny-aware architecture that leverages whole-genome alignments and species trees to model evolutionary relationships explicitly. Trained on alignments spanning vertebrate, mammal and primate evolutionary timescales, GPN-Star achieves state-of-the-art performance across a wide range of variant effect prediction tasks in both coding and non-coding regions of the human genome. Analyses across timescales show task-dependent advantages of modelling more recent versus deeper evolution. To demonstrate its potential to advance human genetics, we show that GPN-Star substantially outperforms previous methods in prioritizing pathogenic and fine-mapped genome-wide association study variants, yields strong enrichments of complex trait heritability and improves power in rare variant association testing. Extending beyond humans, we train GPN-Star for five model organisms—Mus musculus, Gallus gallus, Drosophila melanogaster, Caenorhabditis elegans and Arabidopsis thaliana—demonstrating the robustness and generalizability of the framework. Taken together, these results position GPN-Star as a scalable, powerful and flexible tool for genome interpretation, well suited to leverage the growing abundance of comparative genomics data.
-
Deep Learning-Driven Discovery and Engineering of an Efficient PETase for Depolymerization and Detoxification of PET Microplastics Under Physiological ConditionsYuxuan Wang, Shijie He, Yuheng Chang, Sheng Mao, Jianan Canal Li, and 5 more authorsAdvanced Science, 2026Microplastics (MPs) accumulation in ecosystem and human organs poses urgent environmental and health risks, yet few enzymes efficiently degrade polyethylene terephthalate (PET) under physiological conditions. We leveraged deep learning to mine unexplored sequence space across 246 million proteins, discovering AhPETase, an evolutionarily distinct hydrolase with low homology (<50% sequence identity) to known PET-degrading enzymes. This noncanonical biocatalyst efficiently depolymerizes PET at 37°C, outperforming all typical PETases and achieving a 7.76-fold enhancement over IsPETase, one of the most representative mesophilic PETases. Additionally, engineered variant AhPETaseM1 retains functional activity for over 20 days under physiological conditions and can degrade post-consumer PET MPs 34-fold faster than recombinant human-derived enzyme MG8 (rMG8) under equal enzyme loading. Critically, it reversed PET-induced toxicity in human lung and colon cells, establishing the first proof-of-concept for enzymatic MPs detoxification.
2025
-
A Phylogenetic Approach to Genomic Language ModelingCarlos Albors, Jianan Canal Li, Gonzalo Benegas, Chengzhong Ye, and Yun S SongRECOMB 2025, 2025Genomic language models (gLMs) have shown mostly modest success in identifying evolutionarily constrained elements in mammalian genomes. To address this issue, we introduce a novel framework for training gLMs that explicitly models nucleotide evolution on phylogenetic trees using multispecies whole-genome alignments. Our approach integrates an alignment into the loss function during training but does not require it for making predictions, thereby enhancing the model’s applicability. We applied this framework to train PhyloGPN, a model that excels at predicting functionally disruptive variants from a single sequence alone and demonstrates strong transfer learning capabilities.
-
Large Language Model is Secretly a Protein Sequence OptimizerYinkai Wang, Jiaxing He, Yuanqi Du, Xiaohui Chen, Jianan Canal Li, and 3 more authorsLMRL Workshop at ICLR 2025, 2025We consider the protein sequence engineering problem, which aims to find protein sequences with high fitness levels, starting from a given wild-type sequence. Directed evolution has been a dominating paradigm in this field which has an iterative process to generate variants and select via experimental feedback. We demonstrate large language models (LLMs), despite being trained on massive texts, are secretly protein sequence optimizers. With a directed evolutionary method, LLM can perform protein engineering through Pareto and experiment-budget constrained optimization, demonstrating success on both synthetic and experimental fitness landscapes.
-
Genomic Language Models: Opportunities and ChallengesGonzalo Benegas*, Chengzhong Ye*, Carlos Albors*, Jianan Canal Li*, and Yun S. SongTrends in Genetics, 2025Large language models (LLMs) are having transformative impacts across a wide range of scientific fields, particularly in the biomedical sciences. Just as the goal of Natural Language Processing is to understand sequences of words, a major objective in biology is to understand biological sequences. Genomic Language Models (gLMs), which are LLMs trained on DNA sequences, have the potential to significantly advance our understanding of genomes and how DNA elements at various scales interact to give rise to complex functions. In this review, we showcase this potential by highlighting key applications of gLMs, including fitness prediction, sequence design, and transfer learning. Despite notable recent progress, however, developing effective and efficient gLMs presents numerous challenges, especially for species with large, complex genomes. We discuss major considerations for developing and evaluating gLMs.
2023
-
M²Hub: Unlocking the Potential of Machine Learning for Materials DiscoveryYuanqi Du*, Yingheng Wang*, Yining Huang, Jianan Canal Li, Yanqiao Zhu, and 4 more authorsNeurIPS 2023 Datasets and Benchmarks Track, 2023M2Hub aims to build the machine learning foundations for materials discovery which has a standard workflow from virtual screening/inverse design to simulation to experiment. M2Hub provides data downloading, data processing, (baseline and state-of-the-art) machine learning method implementation, evaluation pipeline and benchmark results.
-
Enzyme Function Prediction using Contrastive LearningTianhao Yu*, Haiyang Cui*, Jianan Canal Li, Yunan Luo, Guangde Jiang, and 1 more authorScience, 2023Enzyme function annotation is a fundamental challenge, and numerous computational tools have been developed. However, most of these tools cannot accurately predict functional annotations, such as enzyme commission (EC) number, for less-studied proteins or those with previously uncharacterized functions or multiple activities. We present a machine learning algorithm named CLEAN (contrastive learning–enabled enzyme annotation) to assign EC numbers to enzymes with better accuracy, reliability, and sensitivity compared with the state-of-the-art tool BLASTp. The contrastive learning framework empowers CLEAN to confidently (i) annotate understudied enzymes, (ii) correct mislabeled enzymes, and (iii) identify promiscuous enzymes with two or more EC numbers—functions that we demonstrate by systematic in silico and in vitro experiments. We anticipate that this tool will be widely used for predicting the functions of uncharacterized enzymes, thereby advancing many fields, such as genomics, synthetic biology, and biocatalysis.
2022
-
MCTensor: A High-Precision Deep Learning Library with Multi-Component Floating-PointTao Yu*, Wentao Guo*, Jianan Canal Li*, Tiancheng Yuan*, and Christopher De SaICML Workshop on Hardware Aware Efficient Training (HAET), 2022In this paper, we introduce MCTensor, a library based on PyTorch for providing general-purpose and high-precision arithmetic for DL training. MCTensor is used in the same way as PyTorch Tensor: we implement multiple basic, matrix-level computation operators and NN modules for MCTensor with identical PyTorch interface. Our algorithms achieve high precision computation and also benefits from heavily-optimized PyTorch floating-point arithmetic. We evaluate MCTensor arithmetic against PyTorch native arithmetic for a series of tasks, where models using MCTensor in float16 would match or outperform the PyTorch model with float32 or float64 precision.
-
Cyclical Kernel Adaptive MetropolisJianan Canal Li*, Yimeng Zeng*, and Wentao Guo*arXiv Preprint, 2022We propose cKAM, cyclical Kernel Adaptive Metropolis, which incorporates a cyclical stepsize scheme to allow control for exploration and sampling. We show that on a crafted bimodal distribution, existing Adaptive Metropolis type algorithms would fail to converge to the true posterior distribution. We point out that this is because adaptive samplers estimates the local/global covariance structure using past history of the chain, which will lead to adaptive algorithms be trapped in a local mode. We demonstrate that cKAM encourages exploration of the posterior distribution and allows the sampler to escape from a local mode, while maintaining the high performance of adaptive methods.