Biology subjects

Nghe, P.

Publications and source records attributed to Nghe, P..

14 recordsLinked to original sources

bioRxiv2026

DORA: a dose-response autoencoder for interpretable transcriptome-to-viability prediction

Predicting the effect of drugs on cell viability is a central challenge in drug discovery. Artificial intelligence holds the promise to considerably accelerate this process by leveraging rich cellular data such as transcriptomics. Current models focus on either transcriptomes or inhibitory concentrations, but they fall short in integrating these sources of information. Here, we propose DORA (Dose-Response Autoencoder), a deep learning model that predicts changes in transcriptomes and viability in a dose-dependent manner, knowing the unperturbed cell state. By enforcing a latent space consistent with cumulative dose effects, DORA matches other methods at predicting transcriptomes and substantially outperforms existing latent representations at viability prediction. The transcriptome-viability relationship provided by the model further allows the recovery of known biomarkers of cell viability while suggesting novel ones. Overall, DORA provides a unified framework delivering actionable biological insights for phenotypic drug screening and personalized medicine.

molecular biology↗

bioRxiv2025

Genipin-Crosslinked, Silane-Anchored 3D Tumor-Stroma Microtissues for High-Content On-Chip Drug Testing

Physiologically relevant 3D tumor models incorporating extracellular matrix (ECM) and cancer-associated fibroblasts (CAFs) are essential for studying tumor progression and drug resistance yet often suffer from hydrogel contraction and instability - especially in microfluidic formats, where ECM deformation hampers long-term culture and quantitative imaging. Here, we present a microfluidic tumor-stroma co-culture platform for head and neck squamous cell carcinoma (HNSCC) that overcomes these limitations through a dual-material strategy: APTES-mediated surface silanization anchors the ECM to the chip, while Genipin-based crosslinking enhances matrix stiffness without compromising cell viability. This approach stabilizes collagen-rich hydrogels for over 10 days, preserving 3D architecture, sustaining >85% viability, and supporting active proliferation. Fourier-transform infrared spectroscopy (FTIR) confirmed successful collagen crosslinking, combining covalent modification of biomaterials with improved mechanical performance. The platform further integrates AI-assisted, high-content imaging to quantify dynamic phenotypic drug responses at both single-cell and higher multicellular/tissue level resolution. Drug chemosensitivity assays, including the co-culture of tumor cells with patient-derived CAFs, demonstrated the quantitative assessment of clinically relevant chemoprotective effects. By combining biomaterial engineering with functional microfluidic design, this system enables reproducible, physiologically relevant modeling of tumor-stroma interactions, offering a scalable tool for preclinical drug screening and personalized medicine or precision oncology applications.

bioengineering↗

bioRxiv2025

Integrating experimental feedback improves generative models for biological sequences

Generative probabilistic models have shown promise in designing artificial RNA and protein sequences but often suffer from high rates of false positives, where sequences predicted as functional fail experimental validation. To address this critical limitation, we explore the impact of reintegrating experimental feedback into the model design process. We propose a likelihood-based reintegration scheme, which we test through extensive computational experiments on both RNA and protein datasets, as well as through wet-lab experiments on the self-splicing ribozyme from the group I intron RNA family where our approach demonstrates particular efficacy. We show that integrating recent experimental data enhances the models capacity of generating functional sequences (e.g. from 6.7% to 63.7% of active designs at 45 mutations). This feedback-driven approach thus provides a significant improvement in the design of biomolecular sequences by directly tackling the false-positive challenge.

bioinformatics↗

bioRxiv2024

A Guide for Active Learning in Synergistic Drug Discovery

Synergistic drug combination screening is a promising strategy in drug discovery, but it involves navigating a costly and complex search space. While AI, particularly deep learning, has advanced synergy predictions, its effectiveness is limited by the low occurrence of synergistic drug pairs. Active learning, which integrates experimental testing into the learning process, has been proposed to address this challenge. In this work, we explore the key components of active learning to provide recommendations for its implementation. We find that molecular encoding has a limited impact on performance, while the cellular environment features significantly enhance predictions. Additionally, active learning can discover 60% of synergistic drug pairs with only exploring 10% of combinatorial space. The synergy yield ratio is observed to be even higher with smaller batch sizes, where dynamic tuning of the exploration-exploitation strategy can further enhance performance.

bioinformatics↗

bioRxiv2024

Expanding the space of self-reproducing RNAs using probabilistic generative models

Estimating the plausibility of RNA self-reproduction is central to origin-of-life scenarios but self-reproduction has been shown in only a handful of systems. Here, we populated a vast sequence space of ribozymes using statistical covariation models and secondary structure prediction. Experimentally assayed sequences were found active as far as 65 mutations from a reference natural sequence. The number of potentially generated sequences together with the experimental success rate indicate that at least [~]1039 such ribozymes may exist. Randomly sampled artificial ribozymes exhibited autocatalytic self-reproduction akin to the reference sequence. The combination of high-throughput screening and probabilistic modeling considerably improves our estimation of the number of self-reproducing systems, paving the way for a statistical approach to the origin of life.

biochemistry↗

bioRxiv2024

Hypothesis-driven interpretable neural network for interactions between genes

Mechanistic models of genetic interactions are rarely feasible due to a lack of information and computational challenges. Alternatively, machine learning (ML) approaches may predict gene interactions if provided with enough data but they lack interpretability. Here, we propose an ML approach for interpretable genotype-to-fitness mapping, the Direct-Latent Interpretable Model (D-LIM). The neural network is built on a strong hypothesis: mutations in different genes cause independent effects in phenotypes, which then interact via non-linear relationships to determine fitness. D-LIM predicts interpretable genotype-to-fitness maps with state-of-the-art accuracy for gene-to-gene and gene-to-environment perturbations in deep mutational scanning of a metabolic pathway, a protein-protein interaction system, and yeast mutants for environmental adaptation. The hypothesis-driven structure of D-LIM offers interpretable features reminiscent of mechanistic models: the inference of phenotypes, identification of trade-offs, and fitness extrapolation outside of the data domain.

bioinformatics↗

bioRxiv2023

Towards Parsimonious Generative Modeling of RNA Families

Generative probabilistic models emerge as a new paradigm in data-driven, evolution-informed design of biomolecular sequences. This paper introduces a novel approach, called Edge Activation Direct Coupling Analysis (eaDCA), tailored to the characteristics of RNA sequences, with a strong emphasis on simplicity, efficiency, and interpretability. eaDCA explicitly constructs sparse coevolutionary models for RNA families, achieving performance levels comparable to more complex methods while utilizing a significantly lower number of parameters. Our approach demonstrates efficiency in generating artificial RNA sequences that closely resemble their natural counterparts in both statistical analyses and SHAPE-MaP experiments, and in predicting the effect of mutations. Notably, eaDCA provides a unique feature: estimating the number of potential functional sequences within a given RNA family. For example, in the case of cyclic di-AMP riboswitches (RF00379), our analysis suggests the existence of approximately 1039 functional nucleotide sequences. While huge compared to the known < 4, 000 natural sequences, this number represents only a tiny fraction of the vast pool of nearly 1082 possible nucleotide sequences of the same length (136 nucleotides). These results underscore the promise of sparse and interpretable generative models, such as eaDCA, in enhancing our understanding of the expansive RNA sequence space.

bioinformatics↗

bioRxiv2023

Adaptation to a phenotypic trade-off requires an interplay between local and global regulators

Organisms depend on their ability to balance multiple phenotypes by evolutionary adaptation. For instance, cellular growth and movement jointly enable critical processes including microbial colonization and cancer invasiveness. Growth and movement are known to be controlled by local regulators that target single operons, as well as by global regulators that impact hundreds of genes. However, how these different levels of regulation interplay during evolution is unclear. Using Escherichia coli growth and motility as a model system, we show that global regulators enable the adaptation of two phenotypes bound by a trade-off, where improvement in one causes deterioration in the other. We measured how CRISPR-mediated knockdowns of global and local transcription factors impact growth and motility in different environments. We find that local regulators mostly modulate motility, while global regulators jointly modulate growth and motility. Genetic perturbations display complex high order interactions between genes and environments. Nevertheless, gene perturbations display consistent patterns in the growth-motility space when grouped by their position in the regulatory hierarchy. These patterns constrain evolutionary scenarios, where local regulators are typically mutated first to optimize motility, then global regulators allow cells to adjust the trade-off between growth and motility. These findings overall highlight the role of pleiotropic regulators for coordinating phenotypic responses in complex environments.

systems biology↗

bioRxiv2019

Darwinian properties and their trade-offs in autocatalytic RNA networks

Discovering autocatalytic chemistries that can evolve is a major goal in systems chemistry and a critical step towards understanding the origin of life. Autocatalytic networks have been discovered in various chemistries, but we lack a general understanding of how network topology controls the Darwinian properties of variation, differential reproduction, and heredity, which are mediated by the chemical composition. Using barcoded sequencing and droplet microfluidics, we establish a landscape of thousands of networks of RNAs that catalyze their own formation from fragments, and derive relationships between network topology and chemical composition. We find that strong variations arise from catalytic innovations perturbing weakly connected networks, and that reproduction increases with global connectivity. These rules imply trade-offs between reproduction and variation, and between compositional persistence and variation along trajectories of network complexification. Overall, connectivity in reaction networks provides a lever to balance variation (to explore chemical states) with reproduction and heredity (persistence being necessary for selection to act), as required for chemical evolution.

evolutionary biology↗

bioRxiv2019

Evolutionary constraints in regulatory networks defined by partial order between phenotypes

Gene regulation networks allow organisms to adapt to diverse environmental niches. However, the constraints underlying the evolution of regulatory phenotypes remain ill-defined both theoretically and experimentally. Here, we show that the concept of partial order identifies such constraints, and test the predictions by experimentally evolving an engineered signal-integrating network in multiple environments. We find that populations: 1) expand in fitness space along the Pareto-optimal front predicted by conflicts in regulatory demands, by fine-tuning binding affinities within the network, 2) expand beyond this constraint by changes in the network structure, thus allowing access to new fitness domains. Strikingly, the constraint predictions are based on whether the network output increases or decreases in response to the different signals, and do not require information on the network architecture or underlying genetics. Overall, our findings show that limited knowledge on current regulatory phenotypes can provide predictions on future evolutionary constraints.

evolutionary biology↗

bioRxiv2019

Metabolic Cost of Rapid Adaptation of Single Yeast Cells Under Unforeseen Challenge

Cells can rapidly adapt to changing environments through non-genetic processes; however, the metabolic cost of such adaptation has never been considered. Here we demonstrate metabolic coupling in a remarkable, rapid adaptation process (10-3cells/hour) by simultaneously measuring metabolism and division of thousands of individual Saccharomyces cerevisiae cells using a droplet microfluidic system. Following a severe challenge, most cells, while not dividing, continue to metabolize, displaying a remarkably wide diversity of metabolic trajectories from which adaptation events can be anticipated. Adaptation requires the consumption of a characteristic amount of energy, indicating that it is an active process. The demonstration that metabolic trajectories predict a priori adaptation events provides the first evidence of tight energetic coupling between metabolism and regulatory reorganization in adaptation.\n\nOne Sentence SummaryDemonstration of the tight coupling between metabolic activity and regulatory processes during rapid adaptation at the single-cell level.

systems biology↗

bioRxiv2019

Transient compartmentalization dynamics in the presence of mutations and noise

We extend a recently introduced framework for transient compartmentalization of replicators with selection dynamics, by including the effect of mutations and noise in such systems. In the presence of mutations, functional replicators (ribozymes) are turned into non-functional ones (parasites). We evaluate the phase diagram of a system undergoing transient compartmentalization with selection. The system can exhibit either coexistence of ribozymes and parasites, or a pure parasite phase. If the mutation rate exceeds a certain level called the error threshold, the only stable phase is the pure parasite one. Transient compartmentalization with selection can relax this error treshold with respect to a bulk quasispecies case, and even allow ribozymes to coexist with faster growing parasites.\n\nIn order to analyze the role of noise, we also introduce a model for the replication of a template by an enzyme. This model admits two regimes: a diffusion limited regime which generates a high noise, and a replication limited regime, which generates a low noise at the population level. Based on this model, we find that, since the ribozyme dynamics belongs to the replication limited regime, the effects of noise on the phase diagram of the system are mostly negligible. Our results underlines the importance of transient compartmentalization for prebiotic scenarios, and may have implications for directed evolution experiments.

biophysics↗

bioRxiv2018

The molecular basis of genetic interaction diversity in a metabolic pathway

Our ability to predict the impact of mutations on traits relevant for disease and evolution remains severely limited by the dependence of their effects on the genetic background and environment. Even when molecular interactions between genes are known, it is unclear how these translate to organism-level interactions between alleles. We therefore characterized the interplay of genetic and environmental dependencies in determining fitness by quantifying ~4,000 fitness interactions between expression variants of two metabolic genes, in different environments. We detect a remarkable variety of environment-dependent interactions, and demonstrate they can be quantitatively explained by a mechanistic model accounting for catabolic flux, metabolite toxicity and expression costs. Complex fitness interactions between mutations can therefore be predicted simply from their simultaneous impact on a few connected molecular phenotypes.

genetics↗

bioRxiv2018

Selection dynamics in transient compartmentalization

Transient compartments have been recently shown to be able to maintain functional replicators in the context of prebiotic studies. Motivated by this experiment, we show that a broad class of selection dynamics is able to achieve this goal. We identify two key parameters, the relative amplification of non-active replicators (parasites) and the size of compartments. Since the basic ingredients of our model are the competition between a host and its parasite, and the diversity generated by small size compartments, our results are relevant to various phage-bacteria or virus-host ecology problems.\n\nPACS numbers: 05.40.-a, 87.14.G-, 87.23.Kg

biophysics↗