Cell states during SARS-CoV-2 infection in human tracheal/bronchial epithelial cells. a) 6162 cells (Fiege et al., 2021) covering samples of mock-infected (0 h), 24 hpi (hours post-infection), and 48 hpi visualized with UMAP. b) Average expression of representative viral genes, IFNs, and antiviral genes in cells within each cluster (state). c) Cell proportions of clusters at different time points (hpi = 0, 24, and 48). The same colors are used for the lines as for the cluster (state) in a). d) Average expression of antiviral genes (IFIT1, IFIT2, IFIT3, IFIT5, IFIH1, OAS1, OAS2, OAS3, OASL, DDX58) in each cell. e) Average expression of viral genes (cov.orf1ab, cov.S, cov.orf3a, cov.orf6, cov.M, cov.N) in each cell. f) Average expression of interferon genes (IFNB1, IFNL1, IFNL2, IFNL3) in each cell. 103 cells (1.7%) are IFN-positive in each cell. g) Progression of viral infection as indicated by changes in cell proportions of different states. Cells are shown separately at each time point in the leftmost column. The right columns show the average expression of antiviral genes, viral genes, and IFNs in each of these cells. Colorkeys indicate the gene expression level from low (white) to high (red, purple, or yellow).

NOVAa model. a) The cell state transitions are included in the NOVAa model. The straight black arrows indicate transitions between cell states. The curved yellow arrows indicate the effects of IFNs on activating antiviral states. The curved purple arrows indicate viral spread to cells with O and a states. The simpler OVA model bypasses the a and N states, by allowing V to induce direct O → A transitions (Fig. S3). b) Final states of a small lattice (50 × 50) simulations at two different values of pa (both at IFN spreading radius R = 1). c) The fraction of cells in each state in the final frozen configuration as a function of pa. A critical transition is observed at pa = pc ∼ 27.8%. At lower values of pa, most cells terminate in the V state, representing an aggressive tissue infection. Simulations were performed on a lattice with linear dimension L = 1000.

Cluster size distribution. a) The distribution P(s) of cluster sizes of infected cells (s = (N + V)⁄L2) for different values of pa, simulated by starting with one infected cell in a 2D square lattice of linear extent L = 2000. b) The exponents from a) are extracted by re-scaling P (s) as shown on the y-axis, yielding τ = 1.83. The cut-off exponent is estimated as σ ∼ 1. Simulations plot the final outbreak sizes from 10,000 initial infections of one cell. The histogram is log-binned with 5 bins per decade. The critical point at pa = pc = 0.28 is determined as the value with the longest scaling regime.

Range of IFN. a) Typical cluster for an R = 1 simulation at pa ∼ pc = 0.278. b) The dependence of pc with R, approximately reproduced by a fit pc ∼ 3−R. For comparison, the OVA model as well as percolation has pc ∼ 3−R⁄2. In all cases, when pa is above pc then the virus is prevented from spreading. c) Cluster distribution for R = 5 at pa ∼ pc = 0.004, at a 5 times larger linear scale than a).