The effectiveness of control measures during the 2022 COVID-19 outbreak in Shanghai, China.

BACKGROUND: In March 2022, the Omicron variant of SARS-CoV-2 spread rapidly in Shanghai, China. The city adopted strict non-pharmacological intervention (NPI) measures, including lockdown (implemented on March 28 in Pudong and April 1 in Puxi) and blanket PCR testing (April 4). This study aims to understand the effect of these measures. METHODS: We tabulated daily case counts from official reports and fitted a two-patch stochastic SEIR model to the data for the period of March 19 to April 21. This model considered two regions in Shanghai, namely Pudong and Puxi, as the implementation of control measures in Shanghai was carried out on different dates in these regions. We verified our fitting results using the data from April 22 to June 26. Finally, we applied the point estimate of parameter values to simulate our model while varying the dates of control measure implementation, and studied the effectiveness of the control measures. RESULTS: Our point estimate for the parameter values yields expected case counts that agree well the data for both the periods from March 19 to April 21 and from April 22 to June 26. Lockdown did not significantly reduce the intra-region transmission rates. Only about 21% cases were reported. The underlying basic reproduction number R0 was 1.7, and the control reproduction number with both lockdown and blanket PCR testing was 1.3. If both measures were implemented on March 19, only about 5.9% infections would be prevented. CONCLUSIONS: Through our analysis, we found that NPI measures implemented in Shanghai were not sufficient to reduce the reproduction number to below unity. Thus, earlier intervention only has limited effect on reducing cases. The outbreak dies out because of only 27% of the population were active in disease transmission, possibly due to a combination of vaccination and lockdown.

Deffini: A family-specific deep neural network model for structure-based virtual screening.

Deep learning-based virtual screening methods have been shown to significantly improve the accuracy of traditional docking-based virtual screening methods. In this paper, we developed Deffini, a structure-based virtual screening neural network model. During training, Deffini learns protein-ligand docking poses to distinguish actives and decoys and then to predict whether a new ligand will bind to the protein target. Deffini outperformed Smina with an average AUC ROC of 0.92 and AUC PRC of 0.44 in 3-fold cross-validation on the benchmark dataset DUD-E. However, when tested on the maximum unbiased validation (MUV) dataset, Deffini achieved poor results with an average AUC ROC of 0.517. We used the family-specific training approach to train the model to improve the model performance and concluded that family-specific models performed better than the pan-family models. To explore the limits of the predictive power of the family-specific models, we constructed Kernie, a new protein kinase dataset consisting of 358 kinases. Deffini trained with the Kernie dataset outperformed all recent benchmarks on the MUV kinases, with an average AUC ROC of 0.745, which highlights the importance of quality datasets in improving the performance of deep neural network models and the importance of using family-specific models.