Segmentation of pneumonia lesions from CT scans of COVID-19 patients is important for accurate diagnosis and follow-up. Deep learning has a potential to automate this task but requires a large set of high-quality annotations that are difficult to collect. Learning from noisy training labels that are easier to obtain has a potential to alleviate this problem. To this end, we propose a novel noise-robust framework to learn from noisy labels for the segmentation task. We first introduce a noise-robust Dice loss that is a generalization of Dice loss for segmentation and Mean Absolute Error (MAE) loss for robustness against noise, then propose a novel COVID-19 Pneumonia Lesion segmentation network (COPLE-Net) to better deal with the lesions with various scales and appearances. The noiserobust Dice loss and COPLE-Net are combined with an adaptive self-ensembling framework for training, where an Exponential Moving Average (EMA) of a student model is used as a teacher model that is adaptively updated by suppressing the contribution of the student to EMA when the student has a large training loss. The student
The rapid spread of coronavirus SARS-CoV-2 greatly threatens global public health but no prophylactic vaccine is available. Here, we report the generation of a replication-incompetent recombinant serotype 5 adenovirus, Ad5-S-nb2, carrying a codon-optimized gene encoding Spike protein (S). In mice and rhesus macaques, intramuscular injection with Ad5-S-nb2 elicits systemic S-specific antibody and cell-mediated immune (CMI) responses. Intranasal inoculation elicits both systemic and pulmonary antibody responses but weaker CMI response. At 30 days after a single vaccination with Ad5-S-nb2 either intramuscularly or intranasally, macaques are protected against SARS-CoV-2 challenge. A subsequent challenge reveals that macaques vaccinated with a 10-fold lower vaccine dosage (1 × 10
10
viral particles) are also protected, demonstrating the effectiveness of Ad5-S-nb2 and the possibility of offering more vaccine dosages within a shorter timeframe. Thus, Ad5-S-nb2 is a promising candidate vaccine and warrants further clinical evaluation.
Monoclonal antibodies (mAbs) encoded by targeting the spike receptor-binding domain (RBD) have been isolated from different COVID-19 patients. However, the existence and prevalence of shared VH3-53-encoded antibodies in the antibody repertoires is not clear. Using antibody repertoire sequencing, we found that the usage of VH3-53 increased after SARS-CoV-2 infection. A highly shared VH3-53-J6 clonotype was identified in 9 out of 13 COVID-19 patients. This clonotype was derived from convergent gene rearrangements with few somatic hypermutations and was evolutionary conserved. We synthesized 34 repertoire-deduced novel VH3-53-J6 heavy chains and paired with a common IGKV1-9 light chain to produce recombinant mAbs. Most of these recombinant mAbs (23/34) possess RBD binding and virus-neutralizing activities, and recognize ACE2 binding site via the same molecular interface. Our computational analysis, validated by laboratory experiments, revealed that VH3-53 antibodies targeting RBD are commonly present in COVID-19 patients' antibody repertoires, indicating many people have germline-like precursor sequences to rapidly generate SARS-CoV-2 neutralizing antibodies. Moreover, antigenspecific mAbs can be digitally obtained through antibody repertoire sequencing and computational analysis.
As an emerging green and sustainable
solvent, a deep eutectic solvent
(DES) applied to the carbon capture process is considered to be a
promising absorbent. This work aims to comprehensively evaluate the
potential and effectiveness of DESs for CO2 capture. First,
a hydrophobic DES, which is composed of tetrabutylammonium bromide
as the hydrogen bond acceptor (HBA) and decanoic acid as the hydrogen
bond donor (HBD) with a molar ratio of 1:2, was screened out from
280 DESs by the conductor-like screening models–segment activity
coefficient (COSMO-SAC) model. Then, quantum chemistry methods were
used to investigate the interaction mechanism between the DES and
CO2. The results show that the interactions between CO2 and the DES are mainly weak hydrogen bonds and van der Waals
dispersion attraction forces. Next, gas–liquid equilibrium
experiments were performed to investigate the effects of temperature
and pressure, the types of HBAs and HBDs, and the molar ratios of
HBA to HBD on the solubility of CO2. The results show that
the process of DES absorbing CO2 obeys Henry’s law
and confirm the reliability of the COSMO-SAC model prediction. Finally,
a rigorous rate-based model for the DES-based postcombustion CO2 capture process was simulated, and the life cycle environmental
sustainability was evaluated and compared with that of the traditional
solvent monoethanolamine, confirming the advantages of the negligible
vapor pressure, thermal stability, and low ecological toxicity of
the DES. This study provides a technical reference for applying new
solvents developed in the laboratory to practical industrial processes.
A comprehensive study of the B cell response against SARS-CoV-2 could be significant for understanding the immune response and developing therapeutical antibodies and vaccines. To define the dynamics and characteristics of the antibody repertoire following SARS-CoV-2 infection, we analyzed the mRNA transcripts of immunoglobulin heavy chain (IgH) repertoires of 24 peripheral blood samples collected between 3 and 111 days after symptom onset from 10 COVID-19 patients. Massive clonal expansion of naïve B cells with limited somatic hypermutation (SHM) was observed in the second week after symptom onset. The proportion of low-SHM IgG clones strongly correlated with spike-specific IgG antibody titers, highlighting the significant activation of naïve B cells in response to a novel virus infection. The antibody isotype switching landscape showed a transient IgA surge in the first week after symptom onset, followed by a sustained IgG elevation that lasted for at least 3 months. SARS-CoV-2 infection elicited poly-germline reactive antibody responses. Interestingly, 17 different IGHV germline genes recombined with IGHJ6 showed significant clonal expansion. By comparing the IgH repertoires that we sequenced with the 774 reported SARS-CoV-2–reactive monoclonal antibodies (mAbs), 13 shared spike-specific IgH clusters were found. These shared spike-specific IgH clusters are derived from the same lineage of several recently published neutralizing mAbs, including CC12.1, CC12.3, C102, REGN10977, and 4A8. Furthermore, identical spike-specific IgH sequences were found in different COVID-19 patients, suggesting a highly convergent antibody response to SARS-CoV-2. Our analysis based on sequencing antibody repertoires from different individuals revealed key signatures of the systemic B cell response induced by SARS-CoV-2 infection.
IMPORTANCE
Although the canonical delineation of serum antibody responses following SARS-CoV-2 infection has been well established, the dynamics of antibody repertoire at the mRNA transcriptional level has not been well understood, especially the correlation between serum antibody titers and the antibody mRNA transcripts. In this study, we analyzed the IgH transcripts and characterized the B cell clonal expansion and differentiation, isotype switching, and somatic hypermutation in COVID-19 patients. This study provided insights at the repertoire level for the B cell response after SARS-CoV-2 infection.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.