Asymmetric catalytic reactions play a critical role in the synthesis of enantiomerically enriched compounds, with broad applications in pharmaceuticals and fine chemicals. While recent advances in machine learning have improved reaction prediction using molecular descriptors and graph-based representations, existing approaches still face challenges such as limited generalizability, poor interpretability, and the scarcity of high-quality, domain-specific datasets. In this work, we propose CHIMERA, a chirality-augmented and molecular interaction-enhanced multi-level representation framework designed to improve both predictive accuracy and interpretability in asymmetric catalysis. CHIMERA integrates molecular fingerprints, graph-based molecular structures, and motif-decomposed graph embeddings within a hierarchical architecture. It incorporates a chirality-aware attention mechanism and a molecular interaction module to highlight atom-level contributions to reaction outcomes. To address data scarcity and distribution shifts, CHIMERA also introduces a domain adaptation module that enhances transferability across reaction types. We evaluate CHIMERA on two curated datasets of oxazoline-catalyzed asymmetric reactions. Experimental results show that CHIMERA outperforms existing baseline models in both predictive performance and interpretability, offering a promising tool for data-driven catalyst design and optimization in asymmetric synthesis.

Molecular representation learning is a fundamental task in computational chemistry and drug discovery. While graph neural networks (GNNs) have significantly advanced molecular representation learning, they rely solely on conventional molecular graph, inherently failing to capture critical three-dimensional electronic structure, which are crucial determinants of molecular interactions. To address this, we propose the Angle-aware Electrostatic Potential point Network (AEP-Net), a novel paradigm that explicitly integrates high-resolution 3D Electrostatic Potential (ESP) represented as point clouds. AEP-Net represents molecules as collections of Atomic-centered Electrostatic-potential Subgraphs (AES). We design an angle-aware convolution mechanism that explicitly captures 3D spatial geometric features through learnable directional bases. Each AES processes its internal ESP points via a field-aware interaction layer, employing a triple weighting mechanism based on angular similarity, spatial proximity, and electrostatic values to capture fine-grained local electrostatic patterns. Features from different AESs are aggregated at the atomic level through global pooling, preserving the integrity of the local ESP environment while enabling a scalable architecture. Subsequently, a hierarchical hybrid network deeply integrates multi-scale features at the point cloud, atomic, and molecular levels,

Mounting evidence underscores the prevalent hierarchical organization of cancer tissues. At the foundation of this hierarchy reside cancer stem cells, a subset of cells endowed with the pivotal role of engendering the entire cancer tissue through cell differentiation. In recent times, substantial attention has been directed toward the phenomenon of cancer cell plasticity, where the dynamic interconversion between cancer stem cells and nonstem cancer cells has garnered significant interest. Since the task of detecting cancer cell plasticity from empirical data remains a formidable challenge, we propose a Bayesian statistical framework designed to infer phenotypic plasticity within cancer cells, utilizing temporal data on cancer stem cell proportions. Our approach is grounded in a stochastic model, adept at capturing the dynamic behaviors of cells. Leveraging Bayesian analysis, we scrutinize the moment equation governing cancer stem cell proportions, derived from the Kolmogorov forward equation of our stochastic model. Our methodology introduces an improved Euler method for parameter estimation within nonlinear ordinary differential equation models, also extending insights to compositional data. Extensive simulations robustly validate the efficacy of our proposed method. To further corroborate our findings, we apply our approach to analyze published data from SW620 colon cancer cell lines. Our results harmonize with in situ experiments, thereby reinforcing the utility of our method in discerning and quantifying phenotypic plasticity within cancer cells.

Hypermutable cancers create opportunities for the development of various immunotherapies, such as immune checkpoint blockade (ICB) therapy. However, emergent studies have revealed that many hypermutated tumors have poor prognosis due to heterogeneous tumor antigen landscapes, yet the underlying mechanisms remain poorly understood. To understand the mechanisms that govern the responses to therapies, we develop mathematical models to explore the impact of combining chemotherapy and ICB therapy on heterogeneous tumors. Our results uncover how chemotherapy reduces antigenic heterogeneity, creating improved immunological conditions within tumors, which, in turn, enhances the therapeutic effect when combined with ICB. Furthermore, our results show that the recovery of the immune system after chemotherapy is crucial for enhancing the response to chemo-ICB combination therapy.

Mitochondrial DNA (mtDNA) variants hold promise as endogenous barcodes for tracking human cell lineages, but their efficacy as reliable lineage markers are hindered by the complex dynamics of mtDNA in somatic tissues. Here, we use computational modeling and single-cell genomics to thoroughly interrogate the origin and clonal dynamics of mtDNA variants across various biological settings. Our findings reveal that the majority of mtDNA variants which are specifically present in a cell subpopulation, termed subpopulation-specific variants, are pre-existing heteroplasmies in the first cell instead of de novo somatic mutations during divisions. Moreover, subpopulation-specific variants demonstrate limited discriminatory power among different genuine lineages under weak clonal expansion; however, certain subpopulation-specific variants with consistently high frequencies among a subpopulation are capable of faithfully labeling cell lineages in scenarios of stringent clonal expansion, such as strongly expanded T cell populations in diseased conditions and clonal hematopoiesis in aged individuals. Inspired by our simulations, we introduce a lineage informative score, facilitating the identification of reliable mitochondrial lineage tracing markers across different modalities of single-cell genomic data. Combining computational modeling and single-cell sequencing, our study reveals that the performance of mitochondrial lineage tracing is highly dependent on the extent of clonal expansion, which thus should be considered when applying mitochondrial lineage tracing.

Tokenization is a critical design choice in genomic language modeling. Widely used schemes---character-level encoding, fixed-length $k$-mers, and greedy subword algorithms such as BPE---show intrinsic limitations on DNA that are magnified by the small four-letter alphabet. To address this, we adapt Ladderpath, an Algorithmic Information Theory method that identifies nested and hierarchical repetitions through optimal information reuse, into a tokenizer tailored for genomic sequences. Integrating this tokenizer into an 86-million-parameter Transformer yields the Ladderpath Tokenized Model (LTM), which surpasses the best existing models---including those several times larger---on 17 of 21 benchmarks. Comparisons with TF-IDF and other frequency-based baselines show that these gains extend beyond simple motif-frequency statistics. LTM's internal representations further exhibit biologically meaningful organization: token embeddings form coherent clusters, and sequence embeddings group promoters, enhancers, and histone-mark-associated regions without task-specific supervision, revealing an emergent structure of functional sequence classes. These findings show that strengthening the information-theoretic basis of tokenization provides a complementary path to architectural innovations and model scaling, enabling more compact and biologically aligned genomic foundation models.

Quantifying dynamic changes in cell populations is crucial for a comprehensive understanding of biological processes such as cell proliferation, injury repair, and disease progression. However, compared to directly measuring the absolute cell numbers of specific subpopulations, relative proportion data demonstrate greater reproducibility and yield more stable, reliable outcomes. Therefore, inferring absolute cell numbers from relative proportion data may present a novel approach for effectively predicting changes in cell population sizes. To address this, we establish two mathematical mappings between cell proportions and population sizes using moment equations derived from stochastic cell-plasticity models. Notably, our findings indicate that one of these mappings does not require prior knowledge of the initial population size, highlighting the value of incorporating variance information into cell proportion data. We evaluated the robustness of our methods from multiple perspectives and extended their application to various biological mechanisms within the context of cell plasticity models. These methods help mitigate the limitations associated with the direct measurement of absolute cell counts through experimental techniques. Moreover, they provide new insights into leveraging the stochastic dynamics of cell populations to quantify interactions between different biomasses within the system.

Tumor-immune interactions are shaped by both antigenic heterogeneity and stochastic perturbations in the tumor microenvironment, yet the mathematical mechanisms underlying immune phase transitions remain poorly understood. We propose a four-compartment dynamical model that incorporates antigen accumulation and immune escape mutations. Bifurcation analysis reveals bistability between immune surveillance and immune escape states, providing a mechanistic explanation for heterogeneous immune outcomes during tumor progression. In the multistable regime, the stable manifold of a saddle point partitions the state space into distinct basins of attraction, determining the long-term fate of the system. We further analyze how stochastic fluctuations in the tumor microenvironment perturb these separatrices, potentially triggering irreversible state transitions. By characterizing the critical noise intensity and estimating the tipping time, we establish a mathematical framework for assessing noise-induced transitions. The model further predicts that increasing tumor cell death can improve system resilience to stochastic perturbations, whereas stronger immune pressure may facilitate immune escape-highlighting the nonlinear and non-monotonic nature of tumor-immune dynamics.

Transition metal dichalcogenide (TMD) monolayers present a singular coupling in their spin and valley degrees of freedom. Moreover, by applying an external magnetic field it is possible to break the energy degeneracy between their K and -K valleys. This valley Zeeman effect opens the possibility of controlling and distinguishing the spin and valley characters of charge carriers in TMDs by their optical transition energies, making these materials promising for the next generation of spintronic and photonic devices. However, the free excitons of pristine TMD monolayers present a moderate valley Zeeman splitting of approximate to 0.23 meV/T. Therefore, alternative excitonic states with higher magnetic responses are mandatory for application purposes. Here, we investigate the magneto-optical properties of ambient exposed WS2 and WSe2 monolayers by circularly polarized magneto-photoluminescence experiments at cryogenic temperatures. A broad lower energy photoluminescence emission related to an ensemble of defects is observed, presenting remarkable valley-related splittings of approximate to 1.45 meV/T and approximate to 1.11 meV/T for WS2 and WSe2 monolayers, respectively. In addition, we report a significant spin polarization of charge carriers in the defect midgap states induced by the external magnetic field. We explain this spin-polarized population and enhanced valley-related splitting in terms of imbalanced spin-flip transitions, leading to a magnetic field-dependent distribution of charge carriers in multiple defect levels. This effect, together with the individual Zeeman shiftings of the midgap states, explains the strong magneto-optical responses observed. Our work uncovers the singular potential of manipulating the light emission of ambient exposed TMD monolayers by an external magnetic field.

Phylodynamic inference (PI) quantifies population dynamics and evolutionary trajectories using phylogenetic trees. Single-cell lineage tracing enables phylogenetic tree reconstruction for thousands of cells in multicellular organisms, facilitating PI at the cellular level. However, cell differentiation and somatic evolution challenge the direct application of existing PI frameworks to somatic tissues. We introduce scPhyloX, a computational framework modeling structured cell populations by leveraging single-cell phylogenetic trees to infer tissue development and tumor evolution dynamics. A key advancement is its ability to infer time-varying parameters, capturing dynamic biological processes. Simulations demonstrate scPhyloX’s accuracy in scenarios including tissue development, disease treatment, and tumor growth. Application to three real datasets reveals insights into somatic dynamics: cycling stem cell overshoot in fly organ development, clonal expansion of multipotent hematopoietic progenitors during human aging, and pronounced subclonal selection in early colorectal tumorigenesis. scPhyloX thus provides a computational approach for investigating somatic tissue development and evolution.

Selective catalysis, particularly when differentiating substrates with similar reactivities in a mixture, is a significant challenge. In this study, anomaly detection algorithms-tools traditionally used for identifying outliers in data cleaning-are applied to catalyst screening. We focus on developing catalytic methods to selectively oxidize cyclic alkanes over linear alkanes in mixtures such as naphtha. By inserting cyclohexane oxidation data one by one into a database of n-hexane oxidization, we used several anomaly detection algorithms to evaluate whether the inserted cyclohexane oxidation data could be considered anomalous. Conditions identified as anomalies imply that they are likely not suitable for n-hexane oxidization. As these anomalies come from conditions for cyclohexane oxidation, they are promising conditions for selective oxidation of cyclohexane while leaving n-hexane unaltered. These anomalies were thus further investigated, leading to the discovery of a specific catalytic approach that selectively oxidizes cyclohexane. This application of anomaly detection offers a novel method to search for selective catalyst for chemical reactions involving mixed substrates.

Investigating the dynamics of growing cell populations is crucial for unraveling key biological mechanisms in living organisms, with many important applications in therapeutics and biochemical engineering. Classical agent-based simulation algorithms are often inefficient for these systems because they track each individual cell, making them impractical for fast (or even exponentially) growing cell populations. To address this challenge, we introduce a novel stochastic simulation approach based on a Feynman-Kac-like representation of the population dynamics. This method, named the Feynman-Kac-inspired Gillespie's Stochastic Simulation Algorithm (FKG-SSA), always employs a fixed number of independently simulated cells for Monte Carlo computation of the system, resulting in a constant computational complexity regardless of the population size. Furthermore, we theoretically show the statistical consistency of the proposed method, indicating its accuracy and reliability. Finally, a couple of biologically relevant numerical examples are presented to illustrate the approach. Overall, the proposed FKG-SSA effectively addresses the challenge of simulating growing cell populations, providing a solid foundation for better analysis of these systems.

Interphases are critical in electrochemical systems, influencing performance by controlling ion transport and stability. This study explores a metal-organic interphase in the electrocatalytic reduction of CO2 (CO2RR) on Cu, extending the concept of interphases to CO2 conversion. Investigating organic modifications on CuOx, we discover metal-organic interphases over 10 nm thick in highly ethanol-selective systems, contrary to the expected monolayer adsorption. Using an automated platform, 1080 CO2RR experiments with 180 molecular modifiers identify functional groups affecting selectivity for ethanol and multi-carbon (C2+) products. We find that these modifiers consistently produce metal-organic interphases on the Cu or CuOx surface. These interphases modulate Cu coordination, CO2RR intermediates, and interfacial water configuration, significantly improving electrocatalytic performance. Testing across 11 CuOx-based catalysts validates this approach, culminating in the development of two electrocatalysts that achieve ~80% faradaic efficiency for C2+ products with ethanol partial current densities up to 328 and 507 mA cm−2. This study highlights the pivotal role of interphases in CO2RR, advancing CO2 conversion technologies. Interphases are crucial in electrochemical systems, but their role in CO2 electroreduction remains underexplored. Here, the authors report their importance in improving electrocatalytic performance by systematically investigating 180 molecular modifiers using an automatic electrocatalysis platform.

The interplay between cellular mechanics and biochemical processes in the cell cycle is not well understood. We propose a quantitative model of cell budding in Saccharomyces cerevisiae as a “weaken-fill-repair” process, linking Newtonian mechanics of the cell wall with biochemical changes that affect its properties. Our model reveals that (1) oscillations in mother cell size during budding are an inevitable outcome of the process; (2) asymmetric division is necessary for the daughter cell to maintain mechanical stiffness; and (3) although various aspects of the cell are constrained and interconnected, the budding process is governed by a single reduced parameter, ψ, which balances osmolyte accumulation with enzymatic wall-weakening to ensure homeostasis. This model provides insights into the evolution of cell walls and their role in cell division, offering a system-level perspective on cell morphology.

Quantifying the size of cell populations is crucial for understanding biological processes such as growth, injury repair, and disease progression. Often, experimental data offer information in the form of relative frequencies of distinct cell types, rather than absolute cell counts. This emphasizes the need to devise effective strategies for estimating absolute cell quantities from fraction data. In response to this challenge, we present two computational approaches grounded in stochastic cell population models: the first-order moment method (FOM) and the second-order moment method (SOM). These methods explicitly establish mathematical mappings from cell fraction to cell population size using moment equations of the stochastic models. Notably, our investigation demonstrates that the SOM method obviates the requirement for a priori knowledge of the initial population size, highlighting the utility of incorporating variance details from cell proportions. The robustness of both the FOM and SOM methods was analyzed from different perspectives. Additionally, we extended the application of the FOM and SOM methods to various biological mechanisms within the context of cell plasticity models. Our methodologies not only assist in mitigating the inherent limitations of experimental techniques when only fraction data is available for detecting cell population size, but they also offer new insights into utilizing the stochastic characteristics of cell population dynamics to quantify interactions between different biomasses within the system.

Two-dimensional dilute magnetic semiconductors has been recently reported in semiconducting transition metal dichalcogenides by the introduction of spin-polarized transition metal atoms as dopants. This is the case of vanadium-doped WS_2 and WSe_2 monolayers, which exhibits a ferromagnetic ordering even above room temperature. However, a broadband characterization of their electronic band structure and its dependence on vanadium concentration is still lacking. Therefore, here we perform power-dependent photoluminescence, resonant four-wave mixing, and differential reflectance spectroscopy to study the optical transitions close to the A exciton energy of vanadium-doped WS_2 monolayers with distinct concentrations. Instead of a single A exciton peak, vanadium-doped samples exhibit two photoluminescence peaks associated with transitions to occupied and unoccupied bands. Moreover, resonant Raman spectroscopy and resonant second-harmonic generation measurements revealed a blueshift in the B exciton but no energy change in the C exciton as vanadium is introduced in the monolayers. Density functional theory calculations showed that the band structure is sensitive to the Hubbard U correction for vanadium and several scenarios are proposed to explain the two photoluminescence peaks around the A exciton energy region. Our work provides the first broadband optical characterization of these two-dimensional dilute magnetic semiconductors, shedding light on the novel electronic features of WS_2 monolayers which are tunable by the vanadium concentration.

Phylodynamics inference (PI) is a powerful approach for quantifying population dynamics and evolutionary trajectories of natural species based on phylogenetic trees. The emergence of single-cell lineage tracing technologies now enables the reconstruction of phylogenetic trees for thousands of individual cells within a multicellular organism, opening avenues for employing PI methodologies at the cellular level. However, the intricate process of cell differentiation poses challenges for directly applying current PI frameworks in somatic tissues. Here, we introduce a novel computational approach called single-cell phylodynamic explorer (scPhyloX), designed to model structured cell populations in various cell states, by leveraging single-cell phylogenetic trees to infer dynamics of tissue development and tumor evolution. Our comprehensive simulations demonstrate the high accuracy of scPhyloX across various biological scenarios. Application of scPhyloX to three real datasets of single-cell lineage tracing unveils novel insights into somatic dynamics, such as the overshoot of cycling stem cell populations in fly organ development, clonal expansion of multipotent progenitors of hematopoiesis during human aging, and pronounced subclonal selection in early colorectal tumorigenesis. Thus, scPhyloX is an innovative computational method for investigating the development and evolution of somatic tissues.

Unravelling the origin and evolution of precancerous lesions is crucial for effectively preventing malignant transformation, yet our current knowledge remains limited1-3. Here we used a base editor-enabled DNA barcoding system4 to comprehensively map single-cell phylogenies in mouse models of intestinal tumorigenesis induced by inflammation or loss of the Apc gene. Through quantitative analysis of high-resolution phylogenies including 260,922 single cells from normal, inflamed and neoplastic intestinal tissues, we identified tens of independent cell lineages undergoing parallel clonal expansions within each lesion. We also found polyclonal origins of human sporadic colorectal polyps through bulk whole-exome sequencing and single-gland whole-genome sequencing. Genomic and clinical data support a model of polyclonal-to-monoclonal transition, with monoclonal lesions representing a more advanced stage. Single-cell RNA sequencing revealed extensive intercellular interactions in early polyclonal lesions, but there was significant loss of interactions during monoclonal transition. Therefore, our data suggest that colorectal precancer is often founded by many different lineages and highlight their cooperative interactions in the earliest stages of cancer formation. These findings provide insights into opportunities for earlier intervention in colorectal cancer.

2D dilute magnetic semiconductors (DMS) based on transition metal dichalcogenides (TMD) offer an innovative pathway for advancing spintronic technologies, including the potential to exploit phenomena such as the valley Zeeman effect. However, the impact of magnetic ordering on the valley degeneracy breaking and on the enhancement of the optical transitions g-factors of these materials remains an open question. Here, a giant effective g-factors ranging between ≈-27 and -69 for the bound exciton at 4 K in vanadium-doped WSe2 monolayers, obtained through magneto-photoluminescence (PL) experiments is reported. This giant g-factor disappears at room temperature, suggesting that this response is associated with a magnetic ordering of the vanadium impurity states at low temperatures. Ab initio calculations for the vanadium-doped WSe2 monolayer confirm the existence of magnetic ordering of the vanadium states, which leads to degeneracy breaking of the valence bands at K and K'. A phenomenological analysis is employed to correlate this splitting with the measured enhanced effective g-factor. The findings shed light on the potential of defect engineering of 2D materials for spintronic applications.

Monolayers of molybdenum disulfide (MoS2) are the most studied two-dimensional (2D) transition-metal dichalcogenides (TMDs), due to its exceptional optical, electronic, and opto-electronic properties. Recent studies have shown the possibility of incorporating a small amount of magnetic transition metals (e.g., Fe, Co, Mn, V) into MoS2 to form a 2D dilute magnetic semiconductor (2D-DMS). However, the origin of the observed ferromagnetism has remained elusive, due to the presence of randomly generated sulfur vacancies during synthesis that can pair with magnetic dopants to form complex dopant-vacancy configurations altering the magnetic order induced by the dopants. By combining high-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) imaging with first-principles density functional theory (DFT) calculations and magnetometry data, we demonstrate the critical effects of sulfur vacancies and their pairings with vanadium atoms on the magnetic ordering in V-doped MoS2 (V-MoS2) monolayers. Additionally, we fabricated a series of field effect transistors on these V-MoS2 monolayers and observed the emergence of p-type behavior as the vanadium concentration increased. Our study sheds light on the origin of ferromagnetism in V-MoS2 monolayers and provides a foundation for future research on defect engineering to tune the electronic and magnetic properties of atomically thin TMD-based DMSs.