Skip to main content

Table 2 Parameter estimates for the best fit description of the T cell proliferation data.

From: Reconstruction of cell population dynamics using CFSE

  

95% confidence intervals

Parameter

QL estimate

From QL alone

Monte Carlo with EM

Monte Carlo with EM + 5% noise

γ 0

0.221

(0.211, 0.230)

(0.203, 0.339)

(0.181, 0.345)

α 0

0.232

(0.222, 0.242)

(0.175, 0.253)

(0.163, 0.283)

γ 1–3

0.419

(0.409, 0.430)

(0.365, 0.443)

(0.356, 0.444)

α 1–3

0.427

(0.412, 0.442)

(0.379, 0.582)

(0.348, 0.595)

γ 4

0.086

(0.077, 0.096)

(0.067, 0.679)

(0.063, 0.478)

α 4

0.340

(0.244, 0.437)

(0.027, 0.691)

(0.094, 0.731)

  1. Subscripts on the QL estimates of parameters refer to generation numbers – e.g., γ 1–3 is the division probability in each 12 h timestep for cells in generations 1–3. We show the 95% confidence intervals obtained (i) using the asymptotic normality of the QL estimator, assuming no uncertainty in cell counts, (ii) by generating Monte Carlo (MC) samples from the original CFSE profiles using the EM method as described in the text, and (iii) using the MC method but also adding 5% noise to total counts as an estimate of error in the estimation of total cell counts.