UVAE: Integration of Heterogeneous Unpaired Data with Imbalanced Classes (preprint)

We introduce the Unbiasing Variational Autoencoder (UVAE), a novel computational framework developed for the integration of unpaired biomedical data streams, with a particular focus on clinical flow cytometry. UVAE effectively addresses the challenges of batch effect correction and data alignment by training a semi-supervised model on partially labeled datasets. This approach enables the simultaneous normalisation and integration of diverse data within a shared latent space. The framework is implemented in Python with a descriptive interface for the specification and incorporation of multiple, partially overlapping data series. UVAE employs a probabilistic model for batch effect normalisation, with a generative capacity for unbiased data reconstruction and inference from heterogeneous samples. Its training process strategically balances class contents during various stages, ensuring accurate representation in statistical analyses. The model's convergence is achieved through a stable, non-adversarial training mechanism, complemented by an automated selection of hyper-parameters via Bayesian optimization. We quantitatively validate the performance of UVAE's constituent components and consequently apply it to the real problem of integrating heterogeneous clinical flow cytometry data collected from COVID-19 patients. We show that the alignment process enhances the statistical signal of cell types associated with severity and enables clustering of subpopulations without the impediment of batch effects. Finally, we demonstrate that homogeneous data generated by UVAE can be used to improve the performance of longitudinal regression for predicting peak disease severity from temporal patient samples.

Sex differences in innate anti-viral immune responses to respiratory viruses (preprint)

Males have excess morbidity and mortality from respiratory viral infections and especially so in COVID-19. The mechanisms explaining this excess in disease burden in males are unknown. Innate immune responses are likely critical in protection against a novel virus like SARS-CoV-2. We hypothesised that innate immune responses may be deficient in males relative to females. To test this we stimulated peripheral blood mononuclear cells (PBMCs) from participants in a population-based birth cohort with three respiratory viruses (rhinoviruses-RV-A16 and RV-A1, and respiratory syncytial virus-RSV) and two viral mimics (R848 and CpG-A, to mimic responses to SARS-CoV-2). We measured interferon (IFN) and IFN-induced chemokine responses and investigated sex differences in virus-induced responses. IFN-α, IFN-β and IFN-γ responses to RV-A16 were deficient in males compared to females, fold-inductions being 1.92-fold ( P< 0.001), 2.04-fold ( P <0.001) and 1.77-fold ( P =0.003) lower in males than females, respectively. Similar significant deficiencies in innate immune responses were observed in males for eleven other cytokine-stimulus pairs. Responses in males were greater than those in females in only one of 35 cytokine-stimulus pairs investigated. Review of healthcare records revealed that 12.1% of males but only 6.6% of females were admitted to hospital with respiratory infections in the first year of life ( P =0.017). Impaired innate anti-viral immunity in males likely results in high morbidity and mortality from respiratory viruses including COVID-19. Males may preferentially benefit from therapies that boost innate anti-viral immune responses. Significance Statement Clinical outcomes including, mortality, Intensive care unit admissions and hospital admissions, during COVID-19 disease are consistently and substantially worse in males than in females. The mechanisms underlying this increased susceptibility to severe disease in males are not understood. We hypothesised that the differential outcomes between sexes could be a consequence of deficient innate interferon responses in males, and more robust innate interferon responses in females. We have investigated such responses in a large population-based cohort and found that indeed males have deficient innate interferon responses to viral stimuli, including stimuli that mimic SARS-CoV-2 infection, relative to females. Our findings have very important treatment implications as interferons are available for clinical use with immediate effect.

Longitudinal immune profiling reveals distinct features of COVID-19 pathogenesis (preprint)

Background The pathogenesis of COVID-19, caused by a novel strain of coronavirus (SARS-CoV-2), involves a complex host-virus interaction and is characterised by an exaggerated immune response, the specific components of which are poorly understood. Here we report the outcome of a longitudinal immune profiling study in hospitalised patients during the peak of the COVID-19 pandemic in the UK and show the relationship between immune responses and severity of the clinical presentation. Methods The Coronavirus Immune Response and Clinical Outcomes (CIRCO) study was conducted at four hospitals in Greater Manchester. Patients with SARS-CoV-2 infection, recruited as close to admission as possible, provided peripheral blood samples at enrolment and sequentially thereafter. Fresh samples were assessed for immune cells and proteins in whole blood and serum. Some samples were also stimulated for 3 hours with LPS and analysed for intracellular proteins. Results were stratified based on patient-level data including severity of symptoms and date of reported symptom onset. Findings Longitudinal analysis showed a very high neutrophil to T cell ratio and abnormal activation of monocytes in the blood, which displayed high levels of the cell cycle marker, Ki67 and low COX-2. These properties all reverted in patient with good outcome. Unexpectedly, multiple aspects of inflammation were diminished as patients progressed in severity and time, even in ITU patients not recovering. Interpretation This is the first detailed longitudinal analysis of COVID-19 patients of varying severity and outcome, revealing common features and aspects that track with severity. Patients destined for a severe outcome can be identified at admission when still displaying mild-moderate symptoms. We provide clues concerning pathogenesis that should influence clinical trials and therapeutics. Targeting pathways involved in neutrophil and monocyte release from the bone marrow should be tested in patients with COVID-19. Funding: The Kennedy Trust for Rheumatology Research, The Wellcome Trust, The Royal Society, The BBSRC, National Institute for Health Research (NIHR) Biomedical Research Centres (BRC).