Evaluation of OCT biomarker changes in treatment-naive neovascular AMD using a deep semantic segmentation algorithm.

OBJECTIVES: To determine real-life quantitative changes in OCT biomarkers in a large set of treatment naive patients in a real-life setting undergoing anti-VEGF therapy. For this purpose, we devised a novel deep learning based semantic segmentation algorithm providing the first benchmark results for automatic segmentation of 11 OCT features including biomarkers for neovascular age-related macular degeneration (nAMD). METHODS: Training of a Deep U-net based semantic segmentation ensemble algorithm for state-of-the-art semantic segmentation performance which was used to analyze OCT features prior to, after 3 and 12 months of anti-VEGF therapy. RESULTS: High F1 scores of almost 1.0 for neurosensory retina and subretinal fluid on a separate hold-out test set with unseen patients. The algorithm performed worse for subretinal hyperreflective material and fibrovascular PED, on par with drusenoid PED, and better in segmenting fibrosis. In the evaluation of treatment naive OCT scans, significant changes occurred for intraretinal fluid (mean: 0.03 µm3 to 0.01 µm3, p < 0.001), subretinal fluid (0.08 µm3 to 0.01 µm3, p < 0.001), subretinal hyperreflective material (0.02 µm3 to 0.01 µm3, p < 0.001), fibrovascular PED (0.12 µm3 to 0.09 µm3, p = 0.02) and central retinal thickness C0 (225.78 µm3 to 169.40 µm3). The amounts of intraretinal fluid, fibrovascular PED, and ERM were predictive of poor outcome. CONCLUSIONS: The segmentation algorithm allows efficient volumetric analysis of OCT scans. Anti-VEGF provokes most potent changes in the first 3 months while a gradual loss of RPE hints at a progressing decline of visual acuity. Additional research is required to understand how these accurate OCT predictions can be leveraged for a personalized therapy regimen.

Reconstructing disease dynamics for mechanistic insights and clinical benefit.

Diseases change over time, both phenotypically and in their underlying molecular processes. Though understanding disease progression dynamics is critical for diagnostics and treatment, capturing these dynamics is difficult due to their complexity and the high heterogeneity in disease development between individuals. We present TimeAx, an algorithm which builds a comparative framework for capturing disease dynamics using high-dimensional, short time-series data. We demonstrate the utility of TimeAx by studying disease progression dynamics for multiple diseases and data types. Notably, for urothelial bladder cancer tumorigenesis, we identify a stromal pro-invasion point on the disease progression axis, characterized by massive immune cell infiltration to the tumor microenvironment and increased mortality. Moreover, the continuous TimeAx model differentiates between early and late tumors within the same tumor subtype, uncovering molecular transitions and potential targetable pathways. Overall, we present a powerful approach for studying disease progression dynamics-providing improved molecular interpretability and clinical benefits for patient stratification and outcome prediction.

Systemic alterations in neutrophils and their precursors in early-stage chronic obstructive pulmonary disease.

Systemic inflammation is established as part of late-stage severe lung disease, but molecular, functional, and phenotypic changes in peripheral immune cells in early disease stages remain ill defined. Chronic obstructive pulmonary disease (COPD) is a major respiratory disease characterized by small-airway inflammation, emphysema, and severe breathing difficulties. Using single-cell analyses we demonstrate that blood neutrophils are already increased in early-stage COPD, and changes in molecular and functional neutrophil states correlate with lung function decline. Assessing neutrophils and their bone marrow precursors in a murine cigarette smoke exposure model identified similar molecular changes in blood neutrophils and precursor populations that also occur in the blood and lung. Our study shows that systemic molecular alterations in neutrophils and their precursors are part of early-stage COPD, a finding to be further explored for potential therapeutic targets and biomarkers for early diagnosis and patient stratification.

Distinct gene programs underpinning disease tolerance and resistance in influenza virus infection.

When challenged with an invading pathogen, the host-defense response is engaged to eliminate the pathogen (resistance) and to maintain health in the presence of the pathogen (disease tolerance). However, the identification of distinct molecular programs underpinning disease tolerance and resistance remained obscure. We exploited transcriptional and physiological monitoring across 33 mouse strains, during in vivo influenza virus infection, to identify two host-defense gene programs-one is associated with hallmarks of disease tolerance and the other with hallmarks of resistance. Both programs constitute generic responses in multiple mouse and human cell types. Our study describes the organizational principles of these programs and validates Arhgdia as a regulator of disease-tolerance states in epithelial cells. We further reveal that the baseline disease-tolerance state in peritoneal macrophages is associated with the pathophysiological response to injury and infection. Our framework provides a paradigm for the understanding of disease tolerance and resistance at the molecular level.

Mature neutrophils and a NF-κB-to-IFN transition determine the unifying disease recovery dynamics in COVID-19.

Disease recovery dynamics are often difficult to assess, as patients display heterogeneous recovery courses. To model recovery dynamics, exemplified by severe COVID-19, we apply a computational scheme on longitudinally sampled blood transcriptomes, generating recovery states, which we then link to cellular and molecular mechanisms, presenting a framework for studying the kinetics of recovery compared with non-recovery over time and long-term effects of the disease. Specifically, a decrease in mature neutrophils is the strongest cellular effect during recovery, with direct implications on disease outcome. Furthermore, we present strong indications for global regulatory changes in gene programs, decoupled from cell compositional changes, including an early rise in T cell activation and differentiation, resulting in immune rebalancing between interferon and NF-κB activity and restoration of cell homeostasis. Overall, we present a clinically relevant computational framework for modeling disease recovery, paving the way for future studies of the recovery dynamics in other diseases and tissues.

An integrative model of cardiometabolic traits identifies two types of metabolic syndrome.

Human diseases arise in a complex ecosystem composed of disease mechanisms and the whole-body state. However, the precise nature of the whole-body state and its relations with disease remain obscure. Here we map similarities among clinical parameters in normal physiological settings, including a large collection of metabolic, hemodynamic, and immune parameters, and then use the mapping to dissect phenotypic states. We find that the whole-body state is faithfully represented by a quantitative two-dimensional model. One component of the whole-body state represents 'metabolic syndrome' (MetS) - a conventional way to determine the cardiometabolic state. The second component is decoupled from the classical MetS, suggesting a novel 'non-classical MetS' that is characterized by dozens of parameters, including dysregulated lipoprotein parameters (e.g. low free cholesterol in small high-density lipoproteins) and attenuated cytokine responses of immune cells to ex vivo stimulations. Both components are associated with disease, but differ in their particular associations, thus opening new avenues for improved personalized diagnosis and treatment. These results provide a practical paradigm to describe whole-body states and to dissect complex disease within the ecosystem of the human body.

Inferring cellular heterogeneity of associations from single cell genomics.

MOTIVATION: Cell-to-cell variation has uncovered associations between cellular phenotypes. However, it remains challenging to address the cellular diversity of such associations. RESULTS: Here, we do not rely on the conventional assumption that the same association holds throughout the entire cell population. Instead, we assume that associations may exist in a certain subset of the cells. We developed CEllular Niche Association (CENA) to reliably predict pairwise associations together with the cell subsets in which the associations are detected. CENA does not rely on predefined subsets but only requires that the cells of each predicted subset would share a certain characteristic state. CENA may therefore reveal dynamic modulation of dependencies along cellular trajectories of temporally evolving states. Using simulated data, we show the advantage of CENA over existing methods and its scalability to a large number of cells. Application of CENA to real biological data demonstrates dynamic changes in associations that would be otherwise masked. AVAILABILITY AND IMPLEMENTATION: CENA is available as an R package at Github: https://github.com/mayalevy/CENA and is accompanied by a complete set of documentations and instructions. CONTACT: iritgv@gmail.com. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Cell composition analysis of bulk genomics using single-cell data.

Single-cell RNA sequencing (scRNA-seq) is a rich resource of cellular heterogeneity, opening new avenues in the study of complex tissues. We introduce Cell Population Mapping (CPM), a deconvolution algorithm in which reference scRNA-seq profiles are leveraged to infer the composition of cell types and states from bulk transcriptome data ('scBio' CRAN R-package). Analysis of individual variations in lungs of influenza-virus-infected mice reveals that the relationship between cell abundance and clinical symptoms is a cell-state-specific property that varies gradually along the continuum of cell-activation states. The gradual change is confirmed in subsequent experiments and is further explained by a mathematical model in which clinical outcomes relate to cell-state dynamics along the activation process. Our results demonstrate the power of CPM in reconstructing the continuous spectrum of cell states within heterogeneous tissues.

Dissection of Influenza Infection In Vivo by Single-Cell RNA Sequencing.

The influenza virus is a major cause of morbidity and mortality worldwide. Yet, both the impact of intracellular viral replication and the variation in host response across different cell types remain uncharacterized. Here we used single-cell RNA sequencing to investigate the heterogeneity in the response of lung tissue cells to in vivo influenza infection. Analysis of viral and host transcriptomes in the same single cell enabled us to resolve the cellular heterogeneity of bystander (exposed but uninfected) as compared with infected cells. We reveal that all major immune and non-immune cell types manifest substantial fractions of infected cells, albeit at low viral transcriptome loads relative to epithelial cells. We show that all cell types respond primarily with a robust generic transcriptional response, and we demonstrate novel markers specific for influenza-infected as opposed to bystander cells. These findings open new avenues for targeted therapy aimed exclusively at infected cells.

ImmQuant: a user-friendly tool for inferring immune cell-type composition from gene-expression data.

: The composition of immune-cell subsets is key to the understanding of major diseases and pathologies. Computational deconvolution methods enable researchers to investigate immune cell quantities in complex tissues based on transcriptome data. Here we present ImmQuant, a software tool allowing immunologists to upload transcription profiles of multiple tissue samples, apply deconvolution methodology to predict differences in cell-type quantities between the samples, and then inspect the inferred cell-type alterations using convenient visualization tools. ImmQuant builds on the DCQ deconvolution algorithm and allows a user-friendly utilization of this method by non-bioinformatician researchers. Specifically, it enables investigation of hundreds of immune cell subsets in mouse tissues, as well as a few dozen cell types in human samples. AVAILABILITY AND IMPLEMENTATION: ImmQuant is available for download at http://csgi.tau.ac.il/ImmQuant/ CONTACT: iritgv@post.tau.ac.ilSupplementary information: Supplementary data are available at Bioinformatics online.

Nucleolin-binding by ErbB2 enhances tumorigenicity of ErbB2-positive breast cancer.

ErbB2 is an important member of the ErbB family, which activates growth and proliferation signaling pathways. ErbB2 is often overexpressed in various malignancies, especially in breast cancer, and is a common target for anti-cancer drugs. Breast cancer is currently one of the leading mortality causes in women, and acquired resistance to ErbB2-targeted therapies is a major obstacle in its treatment. Thus, understanding ErbB2-mediated signaling is crucial for further development of anti-cancer therapeutics and disease treatment. Previously, we have reported that the ErbB receptors interact with the major nucleolar protein nucleolin. In addition to its function in the nucleoli of cells, nucleolin participates in various cellular processes at the cytoplasm and cell-surface. Deregulated nucleolin is frequently overexpressed on the membrane of cancer cells. Here, we show that nucleolin increases colony formation and anchorage-independent growth of ErbB2-overexpressing cells. Importantly, this enhanced tumorigenicity also occurs in human ErbB2-positive breast cancer patients; namely, nucleolin overexpression in these patients is associated with reduced patient survival rates and increased disease-risk. ErbB2-nucleolin complexes are formed endogenously in both normal and cancer cells, and their effect on tumorigenicity is mediated through activation of ErbB2 signaling. Accordingly, nucleolin inhibition reduces cell viability and ErbB2 activation in ErbB2-positive cancer cells.

CoD: inferring immune-cell quantities related to disease states.

MOTIVATION: The immune system comprises a complex network of genes, cells and tissues, coordinated through signaling pathways and cell-cell communications. However, the orchestrated role of the multiple immunological components in disease is still poorly understood. Classifications based on gene-expression data have revealed immune-related signaling pathways in various diseases, but how such pathways describe the immune cellular physiology remains largely unknown. RESULTS: We identify alterations in cell quantities discriminating between disease states using ' Cell type of Disease' (CoD), a classification-based approach that relies on computational immune-cell decomposition in gene-expression datasets. CoD attains significantly higher accuracy than alternative state-of-the-art methods. Our approach is shown to recapitulate and extend previous knowledge acquired with experimental cell-quantification technologies. CONCLUSIONS: The results suggest that CoD can reveal disease-relevant cell types in an unbiased manner, potentially heralding improved diagnostics and treatment. AVAILABILITY AND IMPLEMENTATION: The software described in this article is available at http://www.csgi.tau.ac.il/CoD/.

Linking traits based on their shared molecular mechanisms.

There is growing recognition that co-morbidity and co-occurrence of disease traits are often determined by shared genetic and molecular mechanisms. In most cases, however, the specific mechanisms that lead to such trait-trait relationships are yet unknown. Here we present an analysis of a broad spectrum of behavioral and physiological traits together with gene-expression measurements across genetically diverse mouse strains. We develop an unbiased methodology that constructs potentially overlapping groups of traits and resolves their underlying combination of genetic loci and molecular mechanisms. For example, our method predicts that genetic variation in the Klf7 gene may influence gene transcripts in bone marrow-derived myeloid cells, which in turn affect 17 behavioral traits following morphine injection; this predicted effect of Klf7 is consistent with an in vitro perturbation of Klf7 in bone marrow cells. Our analysis demonstrates the utility of studying hidden causative mechanisms that lead to relationships between complex traits.