NPJ Systems Biology and Applications最新文献

While cancer has traditionally been considered a genetic disease, mounting evidence indicates an important role for non-genetic (epigenetic) mechanisms. Common anti-cancer drugs have recently been observed to induce the adoption of non-genetic drug-tolerant cell states, thereby accelerating the evolution of drug resistance. This confounds conventional high-dose treatment strategies aimed at maximal tumor reduction, since high doses can simultaneously promote non-genetic resistance. In this work, we study optimal dosing of anti-cancer treatment under drug-induced cell plasticity. We show that the optimal dosing strategy steers the tumor to a fixed equilibrium composition between sensitive and tolerant cells, while precisely balancing the trade-off between cell kill and tolerance induction. The optimal equilibrium strategy ranges from applying a low dose continuously to applying the maximum dose intermittently, depending on the dynamics of tolerance induction. We finally discuss how our approach can be integrated with in vitro data to derive patient-specific treatment insights.

The first embryonic division of Caenorhabditis elegans is a model for asymmetric cell division, and identifying the stages of cell division across related species could improve our understanding of the divergence of cellular events and mechanisms. Comparative microscopy of evolutionarily divergent species continues to rely on label-free differential interference contrast (DIC) microscopy due to technical challenges in molecular tagging, with the identification of cell division stages still relying on label-free microscopy. Here, we compare multiple deep convolutional neural networks (CNNs) trained to automate cell stage classification in DIC microscopy movies and interpret the results, with code and classification weights released as OpenSource. The networks are trained to identify if a single frame of a time-series belongs to one of the four morphologically distinct stages: (i) pro-nuclear migration, (ii) centration and rotation, (iii) spindle displacement and (iv) cytokinesis, that had been manually labeled. Three previously described networks, ResNet, VggNet, and EfficientNet, and a customized shallow network, which we refer to as EvoCellNet, achieved 91% or greater accuracy in test data from 23 different nematode species. We find activation vectors of the CNNs of the sparse EvoCellNet correlate with spatial localization of intracellular features of multiple species, such as pro-nuclei, spindle, and spindle-poles. While the pipeline is robust when applied to comparable DIC time-series of C. elegans and C. briggsae embryos, distinct from those on which it was trained and tested, successful classification is limited to images with conserved morphological features. Thus, deep learning networks can be used to generalize the morphological changes across species of nematode embryos, capturing chronology based on low-level intracellular features with biological relevance.

Cancer-associated fibroblasts (CAFs) are key components of the tumor microenvironment (TME). CAF phenotypes are highly heterogeneous and exert anti- and protumorigenic effects. We present a mathematical model that describes cancer-immune-CAF interactions and exploits the heterogeneity of CAF phenotypes to predict cancer progression and treatment response. By simulating multiple treatment options, including targeted monotherapies alone, two different immunotherapies, and a combination of therapies, we have found that CAF composition can impact treatment outcomes, potentially resulting in comparable effectiveness of single-drug treatments and combinatorial approaches or even the ineffectiveness of multicombination therapies. These findings suggest that CAF composition can be a promising indicator, in some cases guiding the choice towards less invasive therapies without compromising effectiveness. Our model indicates that accounting for CAF characteristics might facilitate the matching of targeted treatments, supporting clinical decision-making.

Stromal regulation of cancer dissemination is well recognized, however causal genes remain unidentified. We previously demonstrated that epitheliochorial species have acquired stromal resistance to placental invasion, correlating with reduced rate of cancer malignancies, identifying stromal genes correlating with depth of placental invasion called ELI (Evolved Levels of Invasibility) genes. Similarly, decidualization of human endometrial fibroblasts confers resistance to placental invasion. We hypothesized that both trajectories may convergently use similar pathways, providing an opportunity to identify stromal genes regulating epithelial invasion. We created a gene-set ELI-D1 (ELI-Decidual 1), putatively underlying stromal vulnerability to invasion. ELI-D1 were negatively enriched in T1-T2 stage transition in many human cancers, typically preceding dissemination. We also identified candidate transcriptional regulators underlying variation in ELI-D1 genes across eutherians, functionally showing Nr2f6, and JDP2 can regulate stromal resistance to invasion in human fibroblasts. Our comparative approach provides us with a gene-set linked to stromal vulnerability in human cancers.

The gene regulatory network inference method based on bulk sequencing data not only confuses different types of cells, but also ignores the phenomenon of network dynamic changes with cell state. Single cell transcriptome sequencing technology provides data support for constructing cell type and state specific gene regulatory networks. This study proposes a method for inferring cell type and state specific gene regulatory networks based on scRNA-seq data, called inferCSN. Firstly, inferCSN infers pseudo temporal information from scRNA-seq data and reorders cells based on this information. Because of the uneven distribution of cells in pseudo temporal information, the regulatory relationship tends to lean towards the high-density areas of cells. Therefore, based on the cell state, we divide the cells into different windows to eliminate the temporal information differences caused by cell density. Then, a sparse regression model, combined with reference network information, is used to construct a cell type-specific regulatory network (CSN) for each window. The experimental results on both simulated and real scRNA-seq datasets show that inferCSN outperforms other methods in multiple performance metrics. In addition, experimental results on datasets of different types (such as steady-state and linear datasets) and scales (different cell and gene numbers) show that inferCSN is robust. To further demonstrate the effectiveness and application prospects of inferCSN, we analyzed the gene regulatory network of T cells in different states and different tumor subclons within the tumor microenvironment, and we found that comparing the regulatory networks in different states can reveal immune suppression related signaling pathways.

Hyposmia, a common non-motor symptom in Parkinson's disease (PD) linked to reduced odor sensitivity, is associated with brain structural and functional changes, but dynamic brain activity and altered regional information exchange remain underexplored, limiting insight into underlying brain states. We selected 15 PD patients with severe hyposmia (PD-SH), 15 PD patients with normal cognition (PD-CN), and 15 healthy controls (HC). Using functional MRI, we assessed the brain's spatiotemporal connectivity (brain-state) alterations, and the brain's capacity for higher-order information exchange (synergy and redundancy). A dynamic brain state with complex-long-range connections was significantly reduced in PD-SH and PD-CN, compared to HC. Brain-states consisting of modular-clusters in sensorimotor and frontal areas occurred more frequently in PD-SH than in PD-CN and HC. Higher-order information flow was reduced in PD patients, with PD-SH showing a greater reduction in synergetic information flow in frontal, insula, and left sensory-motor. These findings suggest potential discriminative biomarkers for PD-SH.

We employ a computational framework that integrates mathematical programming and Graph Neural Networks (GNNs) to elucidate functional phenotypic heterogeneity in disease by classifying entire pathways under various conditions of interest. Our approach combines two distinct, yet seamlessly integrated, modeling schemes. First, we leverage Prior Knowledge Networks (PKNs) to reconstruct gene networks from genomic and transcriptomic data. We demonstrate how this can be achieved through mathematical programming optimization and provide examples using comprehensive, established databases. We then tailor GNNs to classify each network as a single data point at graph-level, using various node embeddings and edge attributes. These networks may vary in their biological or molecular annotations, which serve as a labeling scheme for their supervised classification. We apply the framework to the human DNA damage and repair pathway using the TP53 regulon in a pancancer study across cell lines and tumor samples to classify Gene Regulatory Networks (GRNs) across different TP53 mutation types. This approach allows us to identify mutations with distinguishable functional profiles that can be related to specific phenotypes, thus providing a data-driven pipeline for genotype-to-phenotype translation. This scalable approach enables the classification of diverse conditions within the multi-factorial nature of diseases and disentangles their polygenic complexity by revealing new functional patterns through a causal representation.

Cancer cell populations, such as chronic lymphocytic leukemia (CLL), are characterized by aberrant proliferation and plasticity: cells may switch between states so increasing population heterogeneity. Previous works have shown that gene expression noise can impact cell-state transition. To gain better insights into transcriptome-wide expression dynamics and the effect of noise on state transition, here we investigate RNA-Seq data of proliferative (PC) and non-proliferative (NPC) CLL cells. Various data analytics were applied to the whole transcriptome, switch-like toggle (ON/OFF) genes, temporal differentially expressed (DE) genes, and randomly selected genes. Collectively, we identified 2713 temporal DE genes (DESeq2 with 4-fold, p < 0.05) and 1704 toggle genes shaping the differentiation process over a period of 96 h, with 604 overlapping genes between them. Despite their lower numbers compared to DE, toggle genes contributed significantly to gene expression noise in both cell types. Toggle gene analyses revealed the enrichment of genes involved in processes such as G-alpha signaling and muscle contraction as proliferation related RHO-GTPase, interleukin and chemokine signaling, and lymphoid cell communication. Thus, toggle genes, although being random ON/OFF genes, show gene expression functional variability. These results suggest that toggle genes play an important role in shaping cell population plasticity.

Tuberculosis (TB) caused by Mycobacterium tuberculosis (Mtb) kills ~1.3 million people annually. Accordingly, vaccines and sophisticated analytical tools are necessary to evaluate their effectiveness. To address these challenges, we created a Generalized Linear Model (GLM) framework to evaluate high-dimensional flow cytometry data and the multivariable influences on immune responses, accommodating proportional and non-normal data, and violations of assumptions set by classical statistical evaluations. In naïve mice vaccinated with BCG boosted with ID93-GLA-SE, we used GLMs to assess the impact of sex, vaccination, and days post-infection on probabilities of immune cell phenotypes following Mtb challenge. We demonstrate enhanced T cell responses in the lung following BCG + ID93-GLA-SE compared to BCG or ID93-GLA-SE alone, with notable sex differences in humoral immunity. This framework highlights GLMs in assessing complex datasets while enhancing our comprehension of independent continuous and categorical variables on vaccine efficacy, and serves as a foundation for deeper, more complex scenarios.

The immune system's defense abilities rely on the diversity of T and B lymphocytes. T Cell Receptors (TCRs) are generated through V(D)J recombination, where distinct genetic elements combine and undergo modifications, creating extensive variability. In breast cancer, the most frequently diagnosed cancer in women, early detection sometimes helps with highly effective and potentially curative treatment. The TCR repertoire may provide information about tumor status. To test this, we investigated the peripheral blood TCR repertoire and its association with tumor status. We collected blood samples from 98 women, including patients and healthy donors. Following TCR profiling, machine learning of these data was able to show an association between TCR profiles and breast cancer presence or absence with high accuracy (average AUC of 0.96). Our findings imply the immune system retains tumor-relevant, TCR-related, signals detectable in blood. This information could potentially benefit future derivatives from this knowledge, either in the field of detection or treatment.