NPJ Systems Biology and Applications最新文献

Infertility affects millions of couples worldwide, and in vitro fertilization is a key therapeutic strategy for achieving parenthood. Despite advances, the first IVF attempt fails in ~60% of patients, highlighting the need for innovative solutions to improve clinical outcomes. Challenges include the limited ability to study embryo implantation, inadequate methods to test therapeutic drugs, and lack of metrics to evaluate implantation images. To address these issues, we developed ImplantoMetrics, a Fiji plugin for quantitative assessment of trophoblast invasion in combination with a 3D-in-vitro model. ImplantoMetrics uses Convolutional Neural Network and XGBoosting to accurately measure multidimensional expansion patterns. It allows quantitative evaluation of potential therapeutic interventions in vitro and enables a complex study of trophoblast invasion. Compared to manual methods, ImplantoMetrics is ~13-times faster and reduces errors through automation. Beyond implantation research, ImplantoMetrics offers a comprehensive tool to study spheroid invasion in different biological contexts, as e.g. demonstrated here for cancer research.

The interactions between tumor and the immune system are main factors in determining cancer treatment outcomes. In Chronic Myeloid Leukemia (CML), considerable evidence shows that the dynamics between residual leukemia and the patient's immune system can result in either sustained disease control, leading to treatment-free remission (TFR), or disease recurrence. The question remains how to integrate mechanistic and data-driven models to support prediction of treatment outcomes. Starting from classical ecological modeling concepts, which allow to explicitly account for immune interactions at the cellular level, we incorporate time-course data on natural killer (NK) cell number, function, and their tumor-induced suppression into our general model of CML treatment. We identify relevant time scales governing treatment and immune response, enabling refined model calibration using tumor and NK cell time courses from different datasets. While the model successfully describes patient-specific response dynamics, critical parameters for predicting treatment outcome remain uncertain. However, by explicitly incorporating tumor load changes in response to TKI dose alterations, these parameters can be estimated and used to derive model predictions for treatment cessation. Further exploring dynamic changes in the number of functional immune cells, we suggest specific measurement strategies of immune effector cell populations to enhance prediction accuracy for CML recurrence following treatment cessation. The generalizability and flexibility of our approach represent a significant step towards quantitative, personalized medicine that integrates tumor-immune dynamics to guide clinical decisions and optimize dynamic cancer therapies.

Biological phenotypes emerge from complex interactions across molecular layers. Yet, data-driven approaches to infer these regulatory networks have primarily focused on single-omic studies, overlooking inter-layer regulatory relationships. To address these limitations, we developed MINIE, a computational method that integrates multi-omic data from bulk metabolomics and single-cell transcriptomics through a Bayesian regression approach that explicitly models the timescale separation between molecular layers. We validate the method on both simulated datasets and experimental Parkinson's disease data. MINIE exhibits accurate and robust predictive performance across and within omic layers, including curated multi-omic networks and the lac operon. Benchmarking demonstrated significant improvements over state-of-the-art methods while ranking among the top performers in comprehensive single-cell network inference analysis. The integration of regulatory dynamics across molecular layers and temporal scales provides a powerful tool for comprehensive multi-omic network inference.

AMP-activated protein kinase (AMPK) plays a key role in restoring cellular metabolic homeostasis after energy stress. Importantly, AMPK acts as a hub of metabolic signaling, integrating multiple inputs and acting on numerous downstream targets to activate catabolic processes and inhibit anabolic ones. Despite the importance of AMPK signaling, unlike other well-studied pathways, such as MAPK/ERK or NF-κB, only a handful of mechanistic models of AMPK signaling have been developed. Epistemic uncertainty in the biochemical mechanism of AMPK activation, combined with the complexity of the AMPK pathway, makes model development particularly challenging. Here, we leveraged uncertainty quantification (UQ) methods and recently developed AMPK biosensors to construct a new, data-informed model of AMPK signaling. Specifically, we applied Bayesian parameter estimation and model selection to ensure that model predictions and assumptions are well-constrained to measurements of AMPK activity using recently developed AMPK biosensors. As an application of the new model, we predicted AMPK activity in response to exercise-like stimuli. We found that AMPK acts as a time- and exercise-dependent integrator of its input. Our results highlight how UQ can facilitate model development and address epistemic uncertainty in a complex signaling pathway, such as AMPK. This work shows the potential for future applications of UQ in systems biology to drive new biological insights by incorporating state-of-the-art experimental data at all stages of model development.

Elucidating the mechanisms of transport kinetics in plants is crucial to develop crops that can use nutrients efficiently. The plant nitrate transporter NRT1.1 rapidly switches between high- and low-affinity transport modes to maintain an optimal uptake amidst fluctuations in nitrate levels. This functional switch is regulated by NRT1.1 phosphorylation, but the precise mechanisms remain poorly understood. Here, using an integrated molecular and systems-level modeling, we identify mechanisms underlying biphasic behaviour of NRT1.1. Phosphorylation of NRT1.1 and its binding to nitrate impacts its overall flexibility and synergistically modulates its global conformation, impacting the nitrate transport rate. Integrating these observations with a regulatory network involving kinases CIPK8/CIPK23 and calcium binding proteins CBL1/9, reveals that in high nitrate conditions, CIPK8-mediated sequestration of CBL1 disrupts the CIPK23-CBL complex required for NRT1.1 phosphorylation, switching NRT1.1 to a low-affinity mode. Together, our findings untangle the molecular complexity enabling NRT1.1 phosphorylation switch with broader implications in nitrate sensing and molecular-level adaption to fluctuating external nutrient levels.

Cancer cells frequently reprogramme their metabolism to support growth and survival, making metabolic pathways attractive targets for therapy. In this study, we investigated the metabolic effects of three kinase inhibitors and their synergistic combinations in the gastric cancer cell line AGS using genome-scale metabolic models and transcriptomic profiling. We applied the tasks inferred from the differential expression (TIDE) algorithm to infer pathway activity changes in the different conditions. We also explored a variant of TIDE that uses task-essential genes to infer metabolic task changes, providing a complementary perspective to the original algorithm. Our results revealed widespread down-regulation of biosynthetic pathways, particularly in amino acid and nucleotide metabolism. Combinatorial treatments induced condition-specific metabolic alterations, including strong synergistic effects in the PI3Ki-MEKi condition affecting ornithine and polyamine biosynthesis. These metabolic shifts provide insight into drug synergy mechanisms and highlight potential therapeutic vulnerabilities. To support reproducibility, we developed an open-source Python package, MTEApy, implementing both TIDE frameworks.

Digital twins (DTs) can transform randomized clinical trials by improving ethical standards, including safety, informed consent, equity, and data privacy. They also enhance trial efficiency by enabling early detection of adverse events and streamlined design. This paper explores the role of DTs in personalized medicine, from pre-clinical research to post-marketing, while addressing technological, legal, and ethical challenges in implementation.

Antimicrobial resistance (AMR) in Pseudomonas aeruginosa poses a critical global health challenge, with current diagnostics relying on slow, culture-based methods. Here, we present a ML framework leveraging transcriptomic data to predict antibiotic resistance with high accuracy. We applied a genetic algorithm to 414 clinical isolates to identify minimal, highly predictive gene sets (~35-40 genes) distinguishing resistant from susceptible strains for meropenem, ciprofloxacin, tobramycin, and ceftazidime. Automated ML classifiers trained on these sets achieved accuracies of 96-99% on test data (F1 scores: 0.93-0.99), surpassing clinical deployment thresholds. Multiple distinct, non-overlapping gene subsets exhibited comparable performance, suggesting that resistance acquisition is associated with changes in the expression of diverse regulatory and metabolic genes. Comparison with known resistance markers from CARD and operon annotations revealed a substantial number of previously unannotated clusters, highlighting significant knowledge gaps in current AMR understanding. Mapping these genes onto independently modulated gene sets (iModulons) revealed transcriptional adaptations across diverse genetic regions. Overall, this study presents a streamlined machine-learning workflow for transcriptomic data and offers a pathway toward rapid diagnostics and personalized treatment strategies against AMR.

During infection with Plasmodium falciparum in pregnancy, parasites express a unique virulence factor, VAR2CSA, that mediates binding of infected red blood cells to the placenta. A major goal in designing vaccines to protect pregnant women from malaria is to elicit antibodies to VAR2CSA. The challenge is that VAR2CSA is highly polymorphic and identifying conserved epitopes is essential to elicit strain-transcending immunity. Unexpectedly, a mouse monoclonal antibody, 3D10, raised against region II of the unrelated Duffy binding protein from P. vivax (DBPII) cross-reacts with diverse alleles of VAR2CSA in vitro, suggesting that epitopes may be shared across this family of 'Duffy binding-like' (DBL) proteins. Peptide arrays spanning four DBL proteins from two Plasmodium spp, including two alleles of VAR2CSA, DBPII, and PvEBP2 (as a negative control), were screened with 3D10 but the data were too complex to manually identify common epitope sequences. As such, we designed a machine learning framework to analyse the array data. We applied decision trees to extract features correlated to 3D10 binding and evaluated the model on an independent dataset for a rodent Plasmodium DBL protein (PcDBP). Next, we analysed patterns of the features predicted by the model to be strongly associated with 3D10 binding and designed mutant peptides to test complex sequence motifs. Features associated with 3D10 reactivity were mapped onto predicted 3D structures of Plasmodium proteins and validated based on 3D10 reactivity to the recombinant antigens. While the array data identified certain linear epitopes, the framework predicted other epitopes to be conformational. This was demonstrated with PcDBP; as predicted by the model, no linear peptides reacted strongly with 3D10, yet the folded protein was recognized by the antibody in a conformation-dependent manner. With this approach, peptide array data can be mined to extract physicochemical properties of epitopes recognized by cross-reactive antibodies.

We propose a system biology approach to understand how GRNs' dynamical feedback with diffusion of some molecular components underlie the emergence of spatial cellular patterns. We use experimental data on the GRN underlying cell differentiation and spatial arrangement in the root epidermis of WT and mutant Arabidopsis phenotypes to validate our proposal. We test a generalized model of reaction-diffusion, which includes cell-to-cell interaction through lateral inhibition dynamics. The GRN corresponds to the reactive part, and diffusion involves two of its components. The Arabidopsis thaliana root epidermis has a distinct interspersed spatial pattern of hair and non-hair cells. Central to this process is the diffusion of CPC and GL3/EGL3 proteins, which drive lateral inhibition to coordinate cell identity. Existing models have shown a limited predictive power due to incomplete GRN topologies and the lack of explicit diffusion dynamics. Here, we introduce a diffusion-coupled meta-GRN model that integrates positive and negative feedback loops to simulate root epidermal pattern formation in wild-type and mutant lines under varying diffusion levels. By explicitly simulating CPC and GL3/EGL3 protein diffusion, in addition to recovering 28 single and multiple loss-of-function mutant phenotypes, as well as capturing trichoblast and atrichoblast spatial distributions relative to cortex cells, this study presents a 2-D morphospace or phenotypic landscape for epidermis patterning depending on diffusion levels. The findings highlight the critical role of protein diffusion and its dynamic feedback loops with complex GRN in shaping cellular spatial configurations and offer new insights into an extended reaction-diffusion dynamic patterning mechanism that is surely at play in other biological systems.