Supplementary Figure 5: Simulation of the cell-matching process.
From: High-throughput measurement of single-cell growth rates using serial microfluidic mass sensor arrays
Simulation of the cell matching process showing that single cells are reliably matched by our method. (A) We simulate a set of cells sampled from a joint distribution of mass and mass accumulation rate similar to the L1210 cells shown in Figure 2. However, we simulate cells entering the serial SMR array at a rate of 100 cells per hour (two-fold more concentrated than we have used in our experiments). Each cell varies in the time it takes to travel from each cantilever to the next (mean 1.9 minutes, standard deviation 0.3 minutes), and Gaussian noise is added to each buoyant mass measurement (standard deviation 0.05 pg, similar to that of our large-channel device). (B) We then match the measurements in the simulated data. All data points that have been matched together as corresponding to the same cell have been colored the same randomly-chosen color. (C) Comparison of the masses and mass accumulation rates from which the data in (A) was generated, and the observed mass and mass accumulation rates, showing excellent agreement. (D) Comparison of mass accumulation rates from which the data in (A) was generated, and the observed mass accumulation rates, showing excellent agreement, except for in the case of two mismatched cells (off-diagonal points) out of 300 in the simulated dataset.
