ST-former for short-term passenger flow prediction during COVID-19 in urban rail transit system (preprint)

Accurate passenger flow prediction of urban rail transit is essential for improving the performance of intelligent transportation systems, especially during the epidemic. How to dynamically model the complex spatiotemporal dependencies of passenger flow is the main issue in achieving accurate passenger flow prediction during the epidemic. To solve this issue, this paper proposes a brand-new transformer-based architecture called STformer under the encoder-decoder framework specifically for COVID-19. Concretely, we develop a modified self-attention mechanism named Causal-Convolution ProbSparse Self-Attention (CPSA) to model the multiple temporal dependencies of passenger flow with low computational costs. To capture the complex and dynamic spatial dependencies, we introduce a novel Adaptive Multi-Graph Convolution Network (AMGCN) by leveraging multiple graphs in a self-adaptive manner. Additionally, the Multi-source Data Fusion block fuses the passenger flow data, COVID-19 confirmed case data, and the relevant social media data to study the impact of COVID-19 to passenger flow. Experiments on real-world passenger flow datasets demonstrate the superiority of ST-former over the other eleven state-of-the-art methods. Several ablation studies are carried out to verify the effectiveness and reliability of our model structure. Results can provide critical insights for the operation of URT systems.

A high potent synthetic nanobody with broad-spectrum activity neutralizes SARS-Cov-2 virus and Omicron variant through a unique binding mode (preprint)

COVID-19 continues to be a severe public health thread worldwide. The major challenge to control this pandemic is the rapid mutation rate of the SARS-Cov-2 virus, leading to the escape of the protection of vaccines and most of the neutralizing antibodies to date. Thus, it is pivotal to develop neutralizing antibodies with broad-spectrum activity targeting multiple SARS-Cov-2 variants. In this study, we first got a synthetic nanobody (named C5) which could interfere with ACE2 binding to SARS-Cov-2 spike protein and its RBD domain. The affinity matured format of C5 clone, named C5G2, has a single digit nanomolar affinity to RBD domain and inhibit its binding to ACE2 with an IC50 of 3.7 nM. Pseudovirus assay indicated that the monovalent C5G2 could protect the cells from the infection of SARS-Cov-2 wild type virus as well as most of the virus of concern, i.e. Alpha, Beta, Gamma and Omicron variants. Strikingly, C5G2 has the highest potency against omicron among all the variants with the IC50 of 4.9ng/ml (0.3 nM). We further solved the Cryo-EM structure of C5G2 in complex with the Spike trimer. The structure data showed that C5G2 bind to RBD mainly through its CDR3 at a vast region that not overlapping with the ACE2 binding surface. Surprisingly, C5G2 also bind to a distinct epitope residing in the NTD domain of spike protein through the same CDR3 loop, which may further increase its potency against the virus infection. Thus, this bi-paratopic nanobody may be served as an effective drug for the prophylaxis and therapy of the Omicron infection.

Neutralizing breadth of antibodies targeting diverse conserved epitopes between SARS-CoV and SARS-CoV-2 (preprint)

Antibody therapeutics for the treatment of COVID-19 has been highly successful while faces a challenge of the recent emergence of the Omicron variant which escapes the majority of existing SARS-CoV-2 neutralizing antibodies (nAbs). Here, we successfully generated a panel of SARS-CoV-2/SARS-CoV cross-neutralizing antibodies by sequential immunization of the two pseudoviruses. Of which, nAbs X01, X10 and X17 showed broadly neutralizing breadths against most variants of concern (VOCs) and X17 was further identified as a Class 5 nAb with undiminished neutralization against the Omicron variant. Cryo-EM structures of three-antibody in complex with the spike proteins of prototyped SARS-CoV-2, Delta, Omicron and SARS-CoV defined three non-overlapping conserved epitopes on the receptor-binding domain (RBD). The triple antibody cocktail exhibited enhanced resistance to viral escape and effective protection against the infection of Beta variant in hamsters. Our finding will aid the development of both antibody therapeutics and broad vaccines against SARS-CoV-2 and emerging variants.