Genomic epidemiology of early SARS-CoV-2 transmission dynamics in Bangladesh

Abstract

Background: Genomic epidemiology has helped reconstruct the global and regional movement of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). However, there is still a lack of understanding of SARS-CoV-2 spread in some of the world's least developed countries (LDCs).

Methods: To begin to address this disparity, we studied the transmission dynamics of the virus in Bangladesh during the country's first COVID-19 wave by analysing case reports and whole-genome sequences from all eight divisions of the country.

Results: We detected > 50 virus introductions to the country during the period, including during a period of national lockdown. Additionally, through discrete phylogeographic analyses, we identified that geographical distance and population -density and/or -size influenced virus spatial dispersal in Bangladesh.

Conclusions: Overall, this study expands our knowledge of SARS-CoV-2 genomic epidemiology in Bangladesh, shedding light on crucial transmission characteristics within the country, while also acknowledging resemblances and differences to patterns observed in other nations.

Keywords: Bangladesh; Genomic epidemiology; Phylodynamics; SARS-CoV-2.

Fig. 1

Daily COVID-19 confirmed case counts in Bangladesh, 2020. Grey highlighted region indicates the timing of a national lockdown. The x-axis ticks correspond to the start of the named month

Fig. 2

Grey-highlighted regions indicate a national lockdown. (A) Daily COVID-19 confirmed cases in each division. (B) Effective reproduction number (R(t)) by division. Median Rt estimates are shown by black lines, and 95% Bayesian credible intervals by coloured ribbons

Fig. 3

SARS-CoV-2 transmission lineage characteristics in Bangladesh. (A) Number of location state transitions between the phylogenetic traits Bangladesh/Global (imports into Bangladesh = blue, exports from Bangladesh = red), as detected via the robust counting approach. Posterior distributions are truncated at their 95% highest posterior distribution (HPD) interval limits and median estimates are shown using horizontal lines. (B) Duration and timing of the largest Bangladesh transmission lineages (> 10 genomes). Each row represents a transmission lineage, and red dots indicate genome sampling times. Boxes and labels on the right axis show the sampling duration (see Figure S3 for more details on sampling duration per lineage), and number of sampled genomes (n). Asterisks show the median estimated time to most recent common ancestor (TMRCA) for each lineage, with the 95% HPD as a yellow bar (C) Relationship between transmission lineage size and TMRCA, with a dashed line indicating the slope of a linear regression. The Pearson correlation coefficient, 95% confidence interval, and p-value are shown. (D) Partition of Bangladesh genomes into cells representing transmission lineages and singletons, each coloured by estimated duration (time between the lineage’s oldest and most recent genomes). Cell size is proportional to lineage size

Fig. 4

Factors associated with SARS-CoV-2 spread. A) Map of Bangladesh showing the eight geographical groups of districts used as discrete traits in the DTA-GLM. Region centroids are marked by red dots. B and C) Predictors of SARS-CoV-2 spread based on models with either population density (B) or population size (C) included as predictors. Bar and line colours indicate different covariates, with origin and destination predictor of a covariate given the same colour within each plot. Inclusion probability is the posterior expectation that the indicator variable is associated with each predictor E(δ) and suggests that the predictor is associated with different rates of viral diffusion. Bayes Factor (BF) support values for each covariate are indicated by black text annotations. The coefficient (β|δ = 1) represents the contribution of each predictor on a log scale when the predictor is included in the model, with the 95% credible interval of the GLM coefficients (β) represented by horizontal lines

Fig. 5

MCC phylogenies of the two largest lineages detected. (A) Lineage 2, and (B) Lineage 8. Branch lengths represent time, as shown on the axis. Tips are coloured by the sampling geographic region used in the DTA-GLM analyses, as shown in the inset map. Black bars indicate the 95% highest posterior density (HPD) interval for node ages