Biology subjects

Koh, H. Y.

Publications and source records attributed to Koh, H. Y..

7 recordsLinked to original sources

bioRxiv2026

Bioactivity foundation model for hit identification and hit-to-lead optimization from sequence data

Learning from large-scale bioactivity archives is complicated by heterogeneous assay readouts and differences in data reliability. Here we show that a model that predicts probability distributions over binding affinities from sequence data opens model training to vast and diverse troves of heterogeneous bioactivity data. We develop PSICHIC+, a bioactivity foundation model trained on approximately 82 million protein-ligand records using a distributional objective that preserves quantitative resolution while incorporating variation in data reliability. Given a protein sequence and ligand SMILES, PSICHIC+ predicts binding affinity in milliseconds. In zero-shot virtual-screening benchmarks without target-specific training data, it improves enrichment over regression- and classification-based approaches and performs competitively with leading structure-based methods. In lab-in-the-loop campaigns, distribution-guided acquisition improves recovery of high-activity compounds under fixed screening budgets. Combining PSICHIC+ with Boltz-2 improves hit-to-lead affinity prediction over Boltz-2 alone, with mean correlation approaching that of FEP+ on the evaluated benchmarks, without requiring high-resolution protein structures. These results support distributional learning as a scalable approach to using heterogeneous bioactivity data for hit identification and hit-to-lead optimization.

bioinformatics↗

bioRxiv2026

OmniSyn unifies target-aware molecular generation and optimization within a synthesis-native LLM framework across the human proteome

Designing target-specific bioactive molecules with actionable synthesis routes for the human proteome holds enormous potential for expanding therapeutic discovery, but remains a challenge. Existing target-aware generative models often depend on protein structures and generate molecules before assessing synthetic feasibility. Here we present OmniSyn, a protein-sequence-conditioned Mixture-of-Experts (MoE) language model that couples a task-conditioned interaction module with a synthesis-action decoder to generate molecules with explicit synthesis traces, thereby unifying de novo ligand generation, synthesizability projection and hit-to-lead (H2L) optimization within synthesis-traceable chemical space. OmniSyn is pre-trained with self-distillation and post-trained with task-specific reinforcement learning (RL) to adapt expert routing and optimize molecular properties across design modes. On unseen protein targets from MolGenBench, a real-world drug-discovery benchmark, OmniSyn achieves state-of-the-art performance across de novo design and H2L optimization, including target-awareness and hit-rediscovery metrics, despite relying only on protein sequences rather than three-dimensional (3D) pocket structures. By embedding synthesis planning into the design process, OmniSyn shifts molecular generation from a generate-then-filter paradigm toward design-with-synthesis paradigm, transforming virtual predictions into experimentally actionable candidates. Independent AiZynthFinder evaluation yielded retrosynthetic success rates of 68.47% for de novo generation and 71.92% for H2L optimization, improving over the strongest baselines by 61.3% and 184.2%, respectively, and supporting the synthetic feasibility of OmniSyn-generated molecules. In synthesizability projection, OmniSyn further converts outputs from external generative models into close analogues with improved retrosynthetic feasibility while preserving molecular similarity. Having established strong benchmark performance and external retrosynthetic feasibility, we next applied OmniSyn at human-proteome scale, spanning more than 21,000 targets. Rapid sequence-conditioned sampling enabled the construction of, to our knowledge, the largest human-proteome-scale generative virtual library, comprising 2.7 billion target-specific molecules, each accompanied by model-derived synthesis traces and target-specific prioritization scores. By enabling scalable target-specific molecular design with synthesis-aware generation across the human proteome, OmniSyn opens new opportunities for exploring previously inaccessible therapeutic targets, including those lacking experimentally resolved structures.

bioinformatics↗

bioRxiv2026

CAMSAP3 loss of function models suggest causative role in generalized genetic epilepsy

Advancements in next generation sequencing have led to the discovery of hundreds of human epilepsy gene associations. Newly associated genes require functional validation to establish causation and to inform patient treatment in the clinic. A recent exome trio analysis identified predicted pathogenic variants in two patients with generalized epilepsy in the gene CAMSAP3. CAMSAP3 regulates non-centrosomal microtubule dynamics, and the acetylation necessary for normal axonal differentiation and migration. We show that overexpression of patient variants leads to protein degradation and dysregulation of microtubule acetylation in cultured HEK cells. Camsap3 knockout zebrafish also exhibit increased axonal microtubule acetylation as well as epileptic features such as seizure-like swimming behaviors, aberrant inhibitory interneuron development and epileptiform via local field potential. Together these data suggest that CAMSAP3 plays an important role in generalized genetic epilepsy.

neuroscience↗

bioRxiv2025

KcatNet: advancing genome-wide enzyme turnover number prediction through structural enzymatic characterization

Enzyme turnover numbers (Kcat) are fundamental kinetic constants that quantify enzymatic efficiency. Systematic studies of Kcat are essential for characterizing the mechanisms underlying proteomic composition and cellular metabolism. However, experimental measurements of Kcat remain limited and prone to noise. To address this, we present KcatNet, a geometric deep learning model designed for high-throughput prediction of Kcat in metabolic enzymes across all organisms, leveraging paired enzyme sequence and substrate representations. KcatNet consistently outperforms existing predictors, particularly for enzymes with high catalytic efficiency, and demonstrates strong generalization to enzymes that are dissimilar to those in the training set. Furthermore, KcatNet uncovers structural mechanisms and interaction patterns within enzyme-substrate complexes, providing insights into architectural principles that are inaccessible with existing methods by harnessing the representational power of large-scale protein language models. We applied KcatNet to genome-scale Kcat prediction across diverse yeast species, improving proteome allocation predictions by integrating its outputs into metabolic models. Experimental validation further confirmed the models ability to identify enzyme mutants with enhanced activity. By bridging the gap between sequence, structure, and function, KcatNet establishes a robust foundation for advancing understanding of molecular-level mechanisms and accelerating enzyme engineering efforts.

bioinformatics↗

bioRxiv2024

Zebrafish models of candidate human epilepsy-associated genes provide evidence of hyperexcitability

Hundreds of novel candidate human epilepsy-associated genes have been identified thanks to advancements in next-generation sequencing and large genome-wide association studies, but establishing genetic etiology requires functional validation. We generated a list of >2200 candidate epilepsy-associated genes, of which 81 were determined suitable for the generation of loss-of-function zebrafish models via CRISPR/Cas9 gene editing. Of those 81 crispants, 48 were successfully established as stable mutant lines and assessed for seizure-like swim patterns in a primary F2 screen. Evidence of seizure-like behavior was present in 5 (arfgef1, kcnd2, kcnv1, ubr5, wnt8b) of the 48 mutant lines assessed. Further characterization of those 5 lines provided evidence for epileptiform activity via electrophysiology in kcnd2 and wnt8b mutants. Additionally, arfgef1 and wnt8b mutants showed a decrease in the number of inhibitory interneurons in the optic tectum of larval animals. Furthermore, RNAseq revealed convergent transcriptional abnormalities between mutant lines, consistent with their developmental defects and hyperexcitable phenotypes. These zebrafish models provide strongest experimental evidence supporting the role of ARFGEF1, KCND2, and WNT8B in human epilepsy and further demonstrate the utility of this model system for evaluating candidate human epilepsy genes. HighlightsO_LIZebrafish models generated by CRISPR/Cas9 gene editing display seizure-like swim patterns in five candidate human epilepsy genes: arfgef1, kcnd2, kcnv1, ubr5, wnt8b. C_LIO_LILocal field potential abnormalities recorded from kcnd2 and wnt8b crispants provide additional evidence of hyperexcitability. C_LIO_LIArfgef1 and wnt8b mutant larvae have fewer inhibitory interneurons than wild type in the optic tectum. C_LIO_LICRISPR-generated mutants of epilepsy genes displayed convergent transcriptional dysregulation, consistent with developmental abnormalities and their hyperexcitability phenotype. C_LI

genetics↗

bioRxiv2023

Threshold of somatic mosaicism disrupting the brain function

Somatic mosaicism in a fraction of brain cells causes neurodevelopmental disorders, including childhood intractable epilepsy. However, the threshold for somatic mosaicism leading to brain dysfunction is unknown. In this study, we induced various mosaic burdens in mice of focal cortical dysplasia type II (FCD II), featuring mTOR somatic mosaicism and spontaneous behavioral seizures. Mosaic burdens ranged from approximately 1,000 to 40,000 neurons expressing the mTOR mutant in the somatosensory (SSC) or medial prefrontal (PFC) cortex. Surprisingly, just [~]8,000-9,000 neurons expressing the MTOR mutant were sufficient to trigger epileptic seizures. Mutational burden correlated with seizure frequency and onset, with a higher tendency for electrographic inter-ictal spikes and beta- and gamma-frequency oscillations in FCD II mice exceeding the threshold. Moreover, mutation-negative FCD II patients in deep sequencing of their bulky brain tissues revealed somatic mosaicism of mTOR pathway genes as low as 0.07% in resected brain tissues through ultra-deep targeted sequencing (up to 20 million reads). Thus, our study suggests that extremely low levels of somatic mosaicism can contribute to brain dysfunction.

genetics↗

bioRxiv2023

PSICHIC: physicochemical graph neural network for learning protein-ligand interaction fingerprints from sequence data

In drug discovery, determining the binding affinity and functional effects of small-molecule ligands on proteins is critical. Current computational methods can predict these protein-ligand interaction properties but often lose accuracy without high-resolution protein structures and falter in predicting functional effects. We introduce PSICHIC (PhySIcoCHemICal graph neural network), a framework uniquely incorporating physicochemical constraints to decode interaction fingerprints directly from sequence data alone. This enables PSICHIC to attain first-of-its-kind emergent capabilities in deciphering mechanisms underlying protein-ligand interactions, achieving state-of-the-art accuracy and interpretability. Trained on identical protein-ligand pairs without structural data, PSICHIC matched and even surpassed leading structure-based methods in binding affinity prediction. In a library screening for adenosine A1 receptor agonists, PSICHIC discerned functional effects effectively, ranking the sole novel agonist within the top three. PSICHICs interpretable fingerprints identified protein residues and ligand atoms involved in interactions. We foresee PSICHIC reshaping virtual screening and deepening our understanding of protein-ligand interactions.

bioinformatics↗