Prediction of pandemic risk for animal-origin coronavirus using a deep learning method

Background

Coronaviruses can be isolated from bats, civets, pangolins, birds and other wild animals. As an animal-origin pathogen, coronavirus can cross species barrier and cause pandemic in humans. In this study, a deep learning model for early prediction of pandemic risk was proposed based on the sequences of viral genomes.

Methods

A total of 3257 genomes were downloaded from the Coronavirus Genome Resource Library. We present a deep learning model of cross-species coronavirus infection that combines a bidirectional gated recurrent unit network with a one-dimensional convolution. The genome sequence of animal-origin coronavirus was directly input to extract features and predict pandemic risk. The best performances were explored with the use of pre-trained DNA vector and attention mechanism. The area under the receiver operating characteristic curve (AUROC) and the area under precision-recall curve (AUPR) were used to evaluate the predictive models.

Results

The six specific models achieved good performances for the corresponding virus groups (1 for AUROC and 1 for AUPR). The general model with pre-training vector and attention mechanism provided excellent predictions for all virus groups (1 for AUROC and 1 for AUPR) while those without pre-training vector or attention mechanism had obviously reduction of performance (about 5–25%). Re-training experiments showed that the general model has good capabilities of transfer learning (average for six groups: 0.968 for AUROC and 0.942 for AUPR) and should give reasonable prediction for potential pathogen of next pandemic. The artificial negative data with the replacement of the coding region of the spike protein were also predicted correctly (100% accuracy). With the application of the Python programming language, an easy-to-use tool was created to implements our predictor.

Conclusions

Robust deep learning model with pre-training vector and attention mechanism mastered the features from the whole genomes of animal-origin coronaviruses and could predict the risk of cross-species infection for early warning of next pandemic.

Graphical Abstract

Supplementary Information

The online version contains supplementary material available at 10.1186/s40249-021-00912-6.

Using the spike protein feature to predict infection risk and monitor the evolutionary dynamic of coronavirus

BACKGROUND

Coronavirus can cross the species barrier and infect humans with a severe respiratory syndrome. SARS-CoV-2 with potential origin of bat is still circulating in China. In this study, a prediction model is proposed to evaluate the infection risk of non-human-origin coronavirus for early warning.

METHODS

The spike protein sequences of 2666 coronaviruses were collected from 2019 Novel Coronavirus Resource (2019nCoVR) Database of China National Genomics Data Center on Jan 29, 2020. A total of 507 human-origin viruses were regarded as positive samples, whereas 2159 non-human-origin viruses were regarded as negative. To capture the key information of the spike protein, three feature encoding algorithms (amino acid composition, AAC; parallel correlation-based pseudo-amino-acid composition, PC-PseAAC and G-gap dipeptide composition, GGAP) were used to train 41 random forest models. The optimal feature with the best performance was identified by the multidimensional scaling method, which was used to explore the pattern of human coronavirus.

RESULTS

The 10-fold cross-validation results showed that well performance was achieved with the use of the GGAP (g = 3) feature. The predictive model achieved the maximum ACC of 98.18% coupled with the Matthews correlation coefficient (MCC) of 0.9638. Seven clusters for human coronaviruses (229E, NL63, OC43, HKU1, MERS-CoV, SARS-CoV, and SARS-CoV-2) were found. The cluster for SARS-CoV-2 was very close to that for SARS-CoV, which suggests that both of viruses have the same human receptor (angiotensin converting enzyme II). The big gap in the distance curve suggests that the origin of SARS-CoV-2 is not clear and further surveillance in the field should be made continuously. The smooth distance curve for SARS-CoV suggests that its close relatives still exist in nature and public health is challenged as usual.

CONCLUSIONS

The optimal feature (GGAP, g = 3) performed well in terms of predicting infection risk and could be used to explore the evolutionary dynamic in a simple, fast and large-scale manner. The study may be beneficial for the surveillance of the genome mutation of coronavirus in the field.

Inhibitory effect of nicotinamide on in vitro and in vivo production of tumor necrosis factor-alpha

Nicotinamide, a pellagra-preventive factor, has multiple functions such as inhibition of poly-ADP-ribose synthetase, inhibition of inducible nitric oxide synthase, free radical scavenging and suppression of major histocompatibility complex class II expression and ICAM-1 expression on endothelial cells. In addition to these, we have found an inhibitory effect of nicotinamide on production of tumor necrosis factor-alpha (TNF-alpha) in vitro and in vivo. Lipopolysaccharide (LPS)-induced in vitro TNF-alpha production by human peripheral blood mononuclear cells, measured by enzyme-linked immunosorbent assay (ELISA), was significantly inhibited with more than 1 x 10(-3) mol/l of nicotinamide, while interleukin-1-beta was not inhibited and interleukin-6 was slightly inhibited even with 10(-2) mol/l. Oral administration of nicotinamide with more than 62.5 mg/kg also significantly inhibited LPS-induced serum TNF-alpha production measured by ELISA and bioassay in Balb/c mice. Thus, nicotinamide has an inhibitory effect on TNF-alpha production that may be beneficial to TNF-alpha-mediated diseases.

Angiotensin converting enzyme inhibitors suppress production of tumor necrosis factor-alpha in vitro and in vivo

It has been reported that angiotensin converting enzyme (ACE) inhibitors have beneficial effects on insulin resistance and congestive heart failure, in which elevations of serum tumor necrosis factor-alpha (TNF-alpha) level have been indicated. Therefore, in this study, we examined effect of ACE inhibitors on TNF-alpha production both in vitro and in vivo by using human blood mononuclear cells and mice, respectively. LPS (20 micrograms/ml)-induced in vitro TNF-alpha production, measured by bioassay and enzyme-linked immunosorbent assay, was significantly inhibited with captopril, delapril and cilazapril in a concentration of 10(-3) mol/l. A single, oral administration of captopril, delapril and cilazapril at more than 10-fold doses of common clinical use in man significantly inhibited LPS (2 mg/kg)-induced serum TNF-alpha activity in Balb/c mice. These results indicate that ACE inhibitors such as captopril, delapril and cilazapril have an inhibitory effect on TNF-alpha production not only in vitro as previously reported, but also in vivo, although relatively high concentrations and large doses were required in this study.

Effect of long-term treatment with complete Freund's adjuvant on KK-Ay mouse, a model of non-insulin-dependent diabetes mellitus

Nonspecific stimulation with immunoadjuvants significantly improves glucose tolerance in animal models for non-insulin-dependent diabetes mellitus (NIDDM). In this study, we observed the effect of long-term treatment with complete Freund's adjuvant (CFA) on serum factors and histology of various organs in KK-Ay mice with NIDDM. The mice were injected with CFA weekly or every 2 weeks for 12 weeks. Glucose tolerance was significantly improved in the CFA-treated mice throughout the experiment. At the end of the experiment, hypertriglyceridemia was significantly reduced, but serum total protein, glutamic pyruvic transaminase, creatinine, and nonfasting insulin levels were not changed by the treatment. Fatty change of the liver and index of glomerular lesions of the kidney were significantly inhibited in the CFA-treated mice, whereas the pancreatic islet morphology remained unchanged. No toxic effect was observed by the CFA treatment. These results imply that the novel treatment with CFA could control NIDDM and inhibit the development of diabetic glomerular lesions in KK-Ay mice.

Reduced expression of c-Fos in female NOD mouse thymocytes and up-regulation with human lymphotoxin

Previously we reported that the administration of human (h) lymphotoxin (h-LT) markedly protected NOD mice from insulin-dependent diabetes mellitus (IDDM) partly by affecting the generation phase of anti-islet effector cells, probably in the thymus. In this study, we investigated the effect of h-LT on the signal transduction of the mouse thymocytes by observing c-Fos expression in the thymocytes by using a flow cytometer. The intensity of c-Fos expression in whole thymocytes was significantly lower in the female NOD with a high incidence of diabetes than that in the male NOD mice with a low incidence of diabetes and than that in normal mice (P < 0.0001). The low c-Fos expression in the female NOD thymocytes was most prominent in CD3low thymocytes. c-Jun expression of the CD3low thymocytes was also lower in the female NOD mice. Administrations of h-LT, h-TNF, and h-IL-2, which has been reported to prevent IDDM in NOD mice by systemic administration, significantly up-regulated c-Fos expression in CD3low thymocytes. From these results, it is assumed that a relationship may exist between the high diabetes incidence and the defective c-Fos expression in female NOD mice and between the prevention of IDDM and the amelioration of the defective c-Fos expression with h-LT in female NOD mice.