Adaptive immune responses to SARS-CoV-2 persist in the pharyngeal lymphoid tissue of children.

Most studies of adaptive immunity to SARS-CoV-2 infection focus on peripheral blood, which may not fully reflect immune responses at the site of infection. Using samples from 110 children undergoing tonsillectomy and adenoidectomy during the COVID-19 pandemic, we identified 24 samples with evidence of previous SARS-CoV-2 infection, including neutralizing antibodies in serum and SARS-CoV-2-specific germinal center and memory B cells in the tonsils and adenoids. Single-cell B cell receptor (BCR) sequencing indicated virus-specific BCRs were class-switched and somatically hypermutated, with overlapping clones in the two tissues. Expanded T cell clonotypes were found in tonsils, adenoids and blood post-COVID-19, some with CDR3 sequences identical to previously reported SARS-CoV-2-reactive T cell receptors (TCRs). Pharyngeal tissues from COVID-19-convalescent children showed persistent expansion of germinal center and antiviral lymphocyte populations associated with interferon (IFN)-γ-type responses, particularly in the adenoids, and viral RNA in both tissues. Our results provide evidence for persistent tissue-specific immunity to SARS-CoV-2 in the upper respiratory tract of children after infection.

Combination anti-HIV antibodies provide sustained virological suppression.

Antiretroviral therapy is highly effective in suppressing human immunodeficiency virus (HIV)1. However, eradication of the virus in individuals with HIV has not been possible to date2. Given that HIV suppression requires life-long antiretroviral therapy, predominantly on a daily basis, there is a need to develop clinically effective alternatives that use long-acting antiviral agents to inhibit viral replication3. Here we report the results of a two-component clinical trial involving the passive transfer of two HIV-specific broadly neutralizing monoclonal antibodies, 3BNC117 and 10-1074. The first component was a randomized, double-blind, placebo-controlled trial that enrolled participants who initiated antiretroviral therapy during the acute/early phase of HIV infection. The second component was an open-label single-arm trial that enrolled individuals with viraemic control who were naive to antiretroviral therapy. Up to 8 infusions of 3BNC117 and 10-1074, administered over a period of 24 weeks, were well tolerated without any serious adverse events related to the infusions. Compared with the placebo, the combination broadly neutralizing monoclonal antibodies maintained complete suppression of plasma viraemia (for up to 43 weeks) after analytical treatment interruption, provided that no antibody-resistant HIV was detected at the baseline in the study participants. Similarly, potent HIV suppression was seen in the antiretroviral-therapy-naive study participants with viraemia carrying sensitive virus at the baseline. Our data demonstrate that combination therapy with broadly neutralizing monoclonal antibodies can provide long-term virological suppression without antiretroviral therapy in individuals with HIV, and our experience offers guidance for future clinical trials involving next-generation antibodies with long half-lives.

Immune responses to SARS-CoV-2 mRNA vaccination in people with idiopathic CD4 lymphopenia.

BACKGROUND: The immunogenicity of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) mRNA vaccines is variable in individuals with different inborn errors of immunity or acquired immune deficiencies and is yet unknown in people with idiopathic CD4 lymphopenia (ICL). OBJECTIVE: We sought to determine the immunogenicity of mRNA vaccines in patients with ICL with a broad range of CD4 T-cell counts. METHODS: Samples were collected from 25 patients with ICL and 23 age- and sex-matched healthy volunteers (HVs) after their second or third SARS-CoV-2 mRNA vaccine dose. Anti-spike and anti-receptor binding domain antibodies were measured. T-cell receptor sequencing and stimulation assays were performed to quantify SARS-CoV-2-specific T-cell responses. RESULTS: The median age of ICL participants was 51 years, and their median CD4 count was 150 cells/µL; 11 participants had CD4 counts ≤100 cells/µL. Anti-spike IgG antibody levels were greater in HVs than in patients with ICL after 2 and 3 doses of mRNA vaccine. There was no detectable significant difference, however, in anti-S IgG between HVs and participants with ICL and CD4 counts >100 cells/µL. The depth of spike-specific T-cell responses by T-cell receptor sequencing was lower in individuals with ICL. Activation-induced markers and cytokine production of spike-specific CD4 T cells in participants with ICL did not differ significantly compared with HVs after 2 or 3 vaccine doses. CONCLUSIONS: Patients with ICL and CD4 counts >100 cells/µL can mount vigorous humoral and cellular immune responses to SARS-CoV-2 vaccination; however, patients with more severe CD4 lymphopenia have blunted vaccine-induced immunity and may require additional vaccine doses and other risk mitigation strategies.

Perturbations of the T-cell receptor repertoire in response to SARS-CoV-2 in immunocompetent and immunocompromised individuals.

BACKGROUND: Functional T-cell responses are essential for virus clearance and long-term protection after severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection, whereas certain clinical factors, such as older age and immunocompromise, are associated with worse outcome. OBJECTIVE: We sought to study the breadth and magnitude of T-cell responses in patients with coronavirus disease 2019 (COVID-19) and in individuals with inborn errors of immunity (IEIs) who had received COVID-19 mRNA vaccine. METHODS: Using high-throughput sequencing and bioinformatics tools to characterize the T-cell receptor ß repertoire signatures in 540 individuals after SARS-CoV-2 infection, 31 IEI recipients of COVID-19 mRNA vaccine, and healthy controls, we quantified HLA class I- and class II-restricted SARS-CoV-2-specific responses and also identified several HLA allele-clonotype motif associations in patients with COVID-19, including a subcohort of anti-type 1 interferon (IFN-1)-positive patients. RESULTS: Our analysis revealed that elderly patients with COVID-19 with critical disease manifested lower SARS-CoV-2 T-cell clonotype diversity as well as T-cell responses with reduced magnitude, whereas the SARS-CoV-2-specific clonotypes targeted a broad range of HLA class I- and class II-restricted epitopes across the viral proteome. The presence of anti-IFN-I antibodies was associated with certain HLA alleles. Finally, COVID-19 mRNA immunization induced an increase in the breadth of SARS-CoV-2-specific clonotypes in patients with IEIs, including those who had failed to seroconvert. CONCLUSIONS: Elderly individuals have impaired capacity to develop broad and sustained T-cell responses after SARS-CoV-2 infection. Genetic factors may play a role in the production of anti-IFN-1 antibodies. COVID-19 mRNA vaccines are effective in inducing T-cell responses in patients with IEIs.

Immunologic and Virologic Parameters Associated With Human Immunodeficiency Virus (HIV) DNA Reservoir Size in People With HIV Receiving Antiretroviral Therapy.

BACKGROUND: A better understanding of the dynamics of human immunodeficiency virus (HIV) reservoirs in CD4+ T cells of people with HIV (PWH) receiving antiretroviral therapy (ART) is crucial for developing therapies to eradicate the virus. METHODS: We conducted a study involving 28 aviremic PWH receiving ART with high and low levels of HIV DNA. We analyzed immunologic and virologic parameters and their association with the HIV reservoir size. RESULTS: The frequency of CD4+ T cells carrying HIV DNA was associated with higher pre-ART plasma viremia, lower pre-ART CD4+ T-cell counts, and lower pre-ART CD4/CD8 ratios. During ART, the High group maintained elevated levels of intact HIV proviral DNA, cell-associated HIV RNA, and inducible virion-associated HIV RNA. HIV sequence analysis showed no evidence for preferential accumulation of defective proviruses nor higher frequencies of clonal expansion in the High versus Low group. Phenotypic and functional T-cell analyses did not show enhanced immune-mediated virologic control in the Low versus High group. Of considerable interest, pre-ART innate immunity was significantly higher in the Low versus High group. CONCLUSIONS: Our data suggest that innate immunity at the time of ART initiation may play an important role in modulating the dynamics and persistence of viral reservoirs in PWH.

Graft dysfunction in compassionate use of genetically engineered pig-to-human cardiac xenotransplantation: a case report.

BACKGROUND: A genetically engineered pig cardiac xenotransplantation was done on Jan 7, 2022, in a non-ambulatory male patient, aged 57 years, with end-stage heart failure, and on veno-arterial extracorporeal membrane oxygenation support, who was ineligible for an allograft. This report details our current understanding of factors important to the xenotransplantation outcome. METHODS: Physiological and biochemical parameters critical for the care of all heart transplant recipients were collected in extensive clinical monitoring in an intensive care unit. To ascertain the cause of xenograft dysfunction, we did extensive immunological and histopathological studies, including electron microscopy and quantification of porcine cytomegalovirus or porcine roseolovirus (PCMV/PRV) in the xenograft, recipient cells, and tissue by DNA PCR and RNA transcription. We performed intravenous immunoglobulin (IVIG) binding to donor cells and single-cell RNA sequencing of peripheral blood mononuclear cells. FINDINGS: After successful xenotransplantation, the graft functioned well on echocardiography and sustained cardiovascular and other organ systems functions until postoperative day 47 when diastolic heart failure occurred. At postoperative day 50, the endomyocardial biopsy revealed damaged capillaries with interstitial oedema, red cell extravasation, rare thrombotic microangiopathy, and complement deposition. Increased anti-pig xenoantibodies, mainly IgG, were detected after IVIG administration for hypogammaglobulinaemia and during the first plasma exchange. Endomyocardial biopsy on postoperative day 56 showed fibrotic changes consistent with progressive myocardial stiffness. Microbial cell-free DNA testing indicated increasing titres of PCMV/PRV cell-free DNA. Post-mortem single-cell RNA sequencing showed overlapping causes. INTERPRETATION: Hyperacute rejection was avoided. We identified potential mediators of the observed endothelial injury. First, widespread endothelial injury indicates antibody-mediated rejection. Second, IVIG bound strongly to donor endothelium, possibly causing immune activation. Finally, reactivation and replication of latent PCMV/PRV in the xenograft possibly initiated a damaging inflammatory response. The findings point to specific measures to improve xenotransplant outcomes in the future. FUNDING: The University of Maryland School of Medicine, and the University of Maryland Medical Center.

Genome sequence, transcriptome, and annotation of rodent malaria parasite Plasmodium yoelii nigeriensis N67.

BACKGROUND: Rodent malaria parasites are important models for studying host-malaria parasite interactions such as host immune response, mechanisms of parasite evasion of host killing, and vaccine development. One of the rodent malaria parasites is Plasmodium yoelii, and multiple P. yoelii strains or subspecies that cause different disease phenotypes have been widely employed in various studies. The genomes and transcriptomes of several P. yoelii strains have been analyzed and annotated, including the lethal strains of P. y. yoelii YM (or 17XL) and non-lethal strains of P. y. yoelii 17XNL/17X. Genomic DNA sequences and cDNA reads from another subspecies P. y. nigeriensis N67 have been reported for studies of genetic polymorphisms and parasite response to drugs, but its genome has not been assembled and annotated. RESULTS: We performed genome sequencing of the N67 parasite using the PacBio long-read sequencing technology, de novo assembled its genome and transcriptome, and predicted 5383 genes with high overall annotation quality. Comparison of the annotated genome of the N67 parasite with those of YM and 17X parasites revealed a set of genes with N67-specific orthology, expansion of gene families, particularly the homologs of the Plasmodium chabaudi erythrocyte membrane antigen, large numbers of SNPs and indels, and proteins predicted to interact with host immune responses based on their functional domains. CONCLUSIONS: The genomes of N67 and 17X parasites are highly diverse, having approximately one polymorphic site per 50 base pairs of DNA. The annotated N67 genome and transcriptome provide searchable databases for fast retrieval of genes and proteins, which will greatly facilitate our efforts in studying the parasite biology and gene function and in developing effective control measures against malaria.

Type I IFN signaling in T regulatory cells modulates chemokine production and myeloid derived suppressor cells trafficking during EAE.

Interferon-ß has therapeutic efficacy in Multiple Sclerosis by reducing disease exacerbations and delaying relapses. Previous studies have suggested that the effects of type I IFN in Experimental Autoimmune Encephalomyelitis (EAE) in mice were targeted to myeloid cells. We used mice with a conditional deletion (cKO) of the type I IFN receptor (IFNAR) in T regulatory (Treg) cells to dissect the role of IFN signaling on Tregs. cKO mice developed severe EAE with an earlier onset than control mice. Although Treg cells from cKO mice were more activated, the activation status and effector cytokine production of CD4+Foxp3- T cells in the draining lymph nodes (dLN) was similar in WT and cKO mice during the priming phase. Production of chemokines (CCL8, CCL9, CCL22) by CD4+Foxp3- T cells and LN resident cells from cKO mice was suppressed. Suppression of chemokine production was accompanied by a substantial reduction of myeloid derived suppressor cells (MDSCs) in the dLN of cKO mice, while generation of MDSCs and recruitment to peripheral organs was comparable. This study demonstrates that signaling by type I IFNs in Tregs reduces their capacity to suppress chemokine production, with resultant alteration of the entire microenvironment of draining lymph nodes leading to enhancement of MDSC homing, and beneficial effects on disease outcome.

A Mathematical Framework for Understanding Four-Dimensional Heterogeneous Differentiation of CD4+ T Cells.

At least four distinct lineages of CD4+ T cells play diverse roles in the immune system. Both in vivo and in vitro, naïve CD4+ T cells often differentiate into a variety of cellular phenotypes. Previously, we developed a mathematical framework to study heterogeneous differentiation of two lineages governed by a mutual-inhibition motif. To understand heterogeneous differentiation of CD4+ T cells involving more than two lineages, we present here a mathematical framework for the analysis of multiple stable steady states in dynamical systems with multiple state variables interacting through multiple mutual-inhibition loops. A mathematical model for CD4+ T cells based on this framework can reproduce known properties of heterogeneous differentiation involving multiple lineages of this cell differentiation system, such as heterogeneous differentiation of TH1-TH2, TH1-TH17 and iTReg-TH17 under single or mixed types of differentiation stimuli. The model shows that high concentrations of differentiation stimuli favor the formation of phenotypes with co-expression of lineage-specific master regulators.

Co-expression of Foxp3 and Helios facilitates the identification of human T regulatory cells in health and disease.

Foxp3 is regarded as the major transcription factor for T regulatory (Treg) cells and expression of Foxp3 is used to identify and quantitate Treg cells in mouse models. However, several studies have demonstrated that human CD4+ T conventional (Tconv) cells activated in vitro by T cell receptor (TCR) stimulation can express Foxp3. This observation has raised doubt as to the suitability of Foxp3 as a Treg marker in man. Helios, a member of the Ikaros gene family, has been shown to be expressed by 80-90% of human Foxp3+ Treg cells and can potentially serve as a marker of human Treg. Here, we confirm that Foxp3 expression is readily upregulated by Tconv upon TCR stimulation in vitro, while Helios expression is not altered. More importantly, we show that Foxp3 expression is not elevated by stimulation of hTconv in a humanized mouse model of graft versus host disease (GVHD) and in patients with a wide variety of acute and chronic inflammatory diseases including sickle cell disease, acute and chronic GVHD, systemic lupus erythematosus, as well as critical COVID-19. In all patients studied, an excellent correlation was observed between the percentage of CD4+ T cells expressing Foxp3 and the percentage expressing Helios. Taken together, these studies demonstrate that Foxp3 is not induced upon Tconv cell activation in vivo and that Foxp3 expression alone can be used to quantitate Treg cells in humans. Nevertheless, the combined use of Foxp3 and Helios expression provides a more reliable approach for the characterization of Treg in humans.

ImmunoTyper-SR: A Novel Computational Approach for Genotyping Immunoglobulin Heavy Chain Variable Genes using Short Read Data.

Human immunoglobulin heavy chain (IGH) locus on chromosome 14 includes more than 40 functional copies of the variable gene (IGHV), which, together with the joining genes (IGHJ), diversity genes (IGHD), constant genes (IGHC) and immunoglobulin light chains, code for antibodies that identify and neutralize pathogenic invaders as a part of the adaptive immune system. Because of its highly repetitive sequence composition, the IGH locus has been particularly difficult to assemble or genotype through the use of standard short read sequencing technologies. Here we introduce ImmunoTyper-SR, an algorithmic method for genotype and CNV analysis of the germline IGHV genes using Illumina whole genome sequencing (WGS) data. ImmunoTyper-SR is based on a novel combinatorial optimization formulation that aims to minimize the total edit distance between reads and their assigned IGHV alleles from a given database, with constraints on the number and distribution of reads across each called allele. We have validated ImmunoTyper-SR on 12 individuals with Illumina WGS data from the 1000 Genomes Project, whose IGHV allele composition have been studied extensively through the use of long read and targeted sequencing platforms, as well as nine individuals from the NIAID COVID Consortium who have been subjected to WGS twice. We have then applied ImmunoTyper-SR on 585 samples from the NIAID COVID Consortium to investigate associations between distinct IGHV alleles and anti-type I IFN autoantibodies which have been linked to COVID-19 severity.

Deep immune profiling uncovers novel associations with clinical phenotypes of Multisystem Inflammatory Syndrome in Children (MIS-C).

Multisystem Inflammatory Syndrome in Children (MIS-C) is a systemic inflammatory condition that follows SARS-CoV2 infection or exposure in children. Clinical presentations are highly variable and include fever, gastrointestinal (GI) disease, shock, and Kawasaki Disease-like illness (MIS-C/KD). Compared to patients with acute COVID, patients with MIS-C have a distinct immune signature and expansion of TRVB11 expressing T cells. However, the relationship between immunological and clinical phenotypes of MIS-C is unknown. Here, we measured serum biomarkers, TCR repertoire, and SARS-CoV2-specific T cell responses in a cohort of 76 MIS-C patients. Serum biomarkers associated with macrophage and Th1 activation were elevated in patients with shock, consistent with previous reports. Significantly increased SARS-CoV-2-induced IFN-γ, IL-2, and TNF-α production were seen in CD4 + T cells from patients with neurologic involvement and respiratory failure. Diarrhea was associated with a significant reduction in shock-associated serum biomarkers, suggesting a protective effect. TRVB11 usage was highly associated with MIS-C/KD and coronary aneurysms, suggesting a potential biomarker for these manifestations in MIS-C patients. By identifying novel immunologic associations with the different clinical phenotypes of MIS-C, this study provides insights into the clinical heterogeneity of MIS-C. These unique immunophenotypic associations could provide biomarkers to identify patients at risk for severe complications of MIS-C, including shock and MIS-C/KD.

ImmunoTyper-SR: A computational approach for genotyping immunoglobulin heavy chain variable genes using short-read data.

Human immunoglobulin heavy chain (IGH) locus on chromosome 14 includes more than 40 functional copies of the variable gene (IGHV), which are critical for the structure of antibodies that identify and neutralize pathogenic invaders as a part of the adaptive immune system. Because of its highly repetitive sequence composition, the IGH locus has been particularly difficult to assemble or genotype when using standard short-read sequencing technologies. Here, we introduce ImmunoTyper-SR, an algorithmic tool for the genotyping and CNV analysis of the germline IGHV genes on Illumina whole-genome sequencing (WGS) data using a combinatorial optimization formulation that resolves ambiguous read mappings. We have validated ImmunoTyper-SR on 12 individuals, whose IGHV allele composition had been independently validated, as well as concordance between WGS replicates from nine individuals. We then applied ImmunoTyper-SR on 585 COVID patients to investigate the associations between IGHV alleles and anti-type I IFN autoantibodies, which were previously associated with COVID-19 severity.

Immunopathological signatures in multisystem inflammatory syndrome in children and pediatric COVID-19.

Pediatric Coronavirus Disease 2019 (pCOVID-19) is rarely severe; however, a minority of children infected with severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) might develop multisystem inflammatory syndrome in children (MIS-C), with substantial morbidity. In this longitudinal multi-institutional study, we applied multi-omics (analysis of soluble biomarkers, proteomics, single-cell gene expression and immune repertoire analysis) to profile children with COVID-19 (n = 110) and MIS-C (n = 76), along with pediatric healthy controls (pHCs; n = 76). pCOVID-19 was characterized by robust type I interferon (IFN) responses, whereas prominent type II IFN-dependent and NF-κB-dependent signatures, matrisome activation and increased levels of circulating spike protein were detected in MIS-C, with no correlation with SARS-CoV-2 PCR status around the time of admission. Transient expansion of TRBV11-2 T cell clonotypes in MIS-C was associated with signatures of inflammation and T cell activation. The association of MIS-C with the combination of HLA A*02, B*35 and C*04 alleles suggests genetic susceptibility. MIS-C B cells showed higher mutation load than pCOVID-19 and pHC. These results identify distinct immunopathological signatures in pCOVID-19 and MIS-C that might help better define the pathophysiology of these disorders and guide therapy.

Evidence of SARS-CoV-2-Specific T-Cell-Mediated Myocarditis in a MIS-A Case.

A 26-year-old otherwise healthy man died of fulminant myocarditis. Nasopharyngeal specimens collected premortem tested negative for severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). Histopathological evaluation of the heart showed myocardial necrosis surrounded by cytotoxic T-cells and tissue-repair macrophages. Myocardial T-cell receptor (TCR) sequencing revealed hyper-dominant clones with highly similar sequences to TCRs that are specific for SARS-CoV-2 epitopes. SARS-CoV-2 RNA was detected in the gut, supporting a diagnosis of multisystem inflammatory syndrome in adults (MIS-A). Molecular targets of MIS-associated inflammation are not known. Our data indicate that SARS-CoV-2 antigens selected high-frequency T-cell clones that mediated fatal myocarditis.

Quasi-Newton Stochastic Optimization Algorithm for Parameter Estimation of a Stochastic Model of the Budding Yeast Cell Cycle.

Parameter estimation in discrete or continuous deterministic cell cycle models is challenging for several reasons, including the nature of what can be observed, and the accuracy and quantity of those observations. The challenge is even greater for stochastic models, where the number of simulations and amount of empirical data must be even larger to obtain statistically valid parameter estimates. The two main contributions of this work are (1) stochastic model parameter estimation based on directly matching multivariate probability distributions, and (2) a new quasi-Newton algorithm class QNSTOP for stochastic optimization problems. QNSTOP directly uses the random objective function value samples rather than creating ensemble statistics. QNSTOP is used here to directly match empirical and simulated joint probability distributions rather than matching summary statistics. Results are given for a current state-of-the-art stochastic cell cycle model of budding yeast, whose predictions match well some summary statistics and one-dimensional distributions from empirical data, but do not match well the empirical joint distributions. The nature of the mismatch provides insight into the weakness in the stochastic model.

Molecular Network Dynamics of Cell Cycle Control: Periodicity of Start and Finish.

The cell division cycle is controlled by a complex regulatory network which ensures that the phases of the cell cycle are executed in the right order. This regulatory network receives signals from the environment, monitors the state of the DNA, and decides timings of cell cycle events. The underlying transcriptional and post-translational regulatory interactions lead to complex dynamical responses, such as the oscillations in the levels of cell cycle proteins driven by intertwined biochemical reactions. A cell moves between different phases of its cycle similar to a dynamical system switching between its steady states. The complex molecular network driving these phases has been investigated in previous computational systems biology studies. Here, we review the critical physiological and molecular transitions that occur in the cell cycle and discuss the role of mathematical modeling in elucidating these transitions and understand cell cycle synchronization.

Predicting network modules of cell cycle regulators using relative protein abundance statistics.

BACKGROUND: Parameter estimation in systems biology is typically done by enforcing experimental observations through an objective function as the parameter space of a model is explored by numerical simulations. Past studies have shown that one usually finds a set of "feasible" parameter vectors that fit the available experimental data equally well, and that these alternative vectors can make different predictions under novel experimental conditions. In this study, we characterize the feasible region of a complex model of the budding yeast cell cycle under a large set of discrete experimental constraints in order to test whether the statistical features of relative protein abundance predictions are influenced by the topology of the cell cycle regulatory network. RESULTS: Using differential evolution, we generate an ensemble of feasible parameter vectors that reproduce the phenotypes (viable or inviable) of wild-type yeast cells and 110 mutant strains. We use this ensemble to predict the phenotypes of 129 mutant strains for which experimental data is not available. We identify 86 novel mutants that are predicted to be viable and then rank the cell cycle proteins in terms of their contributions to cumulative variability of relative protein abundance predictions. Proteins involved in "regulation of cell size" and "regulation of G1/S transition" contribute most to predictive variability, whereas proteins involved in "positive regulation of transcription involved in exit from mitosis," "mitotic spindle assembly checkpoint" and "negative regulation of cyclin-dependent protein kinase by cyclin degradation" contribute the least. These results suggest that the statistics of these predictions may be generating patterns specific to individual network modules (START, S/G2/M, and EXIT). To test this hypothesis, we develop random forest models for predicting the network modules of cell cycle regulators using relative abundance statistics as model inputs. Predictive performance is assessed by the areas under receiver operating characteristics curves (AUC). Our models generate an AUC range of 0.83-0.87 as opposed to randomized models with AUC values around 0.50. CONCLUSIONS: By using differential evolution and random forest modeling, we show that the model prediction statistics generate distinct network module-specific patterns within the cell cycle network.

Genotype-driven identification of a molecular network predictive of advanced coronary calcium in ClinSeq® and Framingham Heart Study cohorts.

BACKGROUND: One goal of personalized medicine is leveraging the emerging tools of data science to guide medical decision-making. Achieving this using disparate data sources is most daunting for polygenic traits. To this end, we employed random forests (RFs) and neural networks (NNs) for predictive modeling of coronary artery calcium (CAC), which is an intermediate endo-phenotype of coronary artery disease (CAD). METHODS: Model inputs were derived from advanced cases in the ClinSeq®; discovery cohort (n=16) and the FHS replication cohort (n=36) from 89 th -99 th CAC score percentile range, and age-matched controls (ClinSeq®; n=16, FHS n=36) with no detectable CAC (all subjects were Caucasian males). These inputs included clinical variables and genotypes of 56 single nucleotide polymorphisms (SNPs) ranked highest in terms of their nominal correlation with the advanced CAC state in the discovery cohort. Predictive performance was assessed by computing the areas under receiver operating characteristic curves (ROC-AUC). RESULTS: RF models trained and tested with clinical variables generated ROC-AUC values of 0.69 and 0.61 in the discovery and replication cohorts, respectively. In contrast, in both cohorts, the set of SNPs derived from the discovery cohort were highly predictive (ROC-AUC ≥0.85) with no significant change in predictive performance upon integration of clinical and genotype variables. Using the 21 SNPs that produced optimal predictive performance in both cohorts, we developed NN models trained with ClinSeq®; data and tested with FHS data and obtained high predictive accuracy (ROC-AUC=0.80-0.85) with several topologies. Several CAD and "vascular aging" related biological processes were enriched in the network of genes constructed from the predictive SNPs. CONCLUSIONS: We identified a molecular network predictive of advanced coronary calcium using genotype data from ClinSeq®; and FHS cohorts. Our results illustrate that machine learning tools, which utilize complex interactions between disease predictors intrinsic to the pathogenesis of polygenic disorders, hold promise for deriving predictive disease models and networks.