Table 2.
Comparing alternative models which assume different contributions of Py-specific (PyTCR) and non-specific (OT1) T cells to cluster formation.
| Model | λ0, 1/h | λ1, 1/h | AIC | w | |
|---|---|---|---|---|---|
| Only PyTCR cells recruit | 0.14 (0.1, 0.19) | 0.58 (0.46, 0.74) | 128.0 | 260.3 | 0.40 |
| Only OT1 cells recruit | 0.15 (0.11, 0.21) | 0.60 (0.47, 0.76) | 131.4 | 267.8 | 0.01 |
| PyTCR and OT1 cells recruit at the same rate | 0.12 (0.085, 0.17) | 0.32 (0.26, 0.39) | 130.7 | 265.6 | 0.03 |
| PyTCR and OT1 cells recruit at different rates | 0.17 (0.13, 0.23) | PyTCR = 0.67 (0.54, 0.85), OT1 = −0.16 (−0.03, −0.22) | 127.1 | 260.6 | 0.34 |
| PyTCR and OT1 cells recruit at different rates toward different cell types | 0.18 (0.13, 0.24) | PyTCR:PyTCR = 0.73 (0.58, 0.91), PyTCR:OT1 = 0.61 (0.43, 0.84), OT1:OT1 = −0.15 (−0.05, −0.37), OT1:PyTCR = −0.23 (−0.02, −0.22) | 126.7 | 264.8 | 0.04 |
We fit the basic mathematical model on co-clustering of Plasmodium-specific and non-specific T cells (Equations 12–15) to the data on co-clustering of T cells around Py liver stages assuming DD recruitment model and different mechanisms of how T cells contribute to cluster formation (see Table 1 for tested models). Here we list the estimated initial recruitment rate λ0 and how recruitment rate changes with cluster size λ1 (i.e., in the DD recruitment model the recruitment rate is λk = λ0+kλ1), the negative log-likelihood , AIC, and Akaike weights w for the model fit. In these fits the total exit rate of T cells from the cluster of size k was fixed to μk = 0.5k/h. In the column with estimates for λ1 we list specifically the predicted change in the cluster “attractiveness” by a given type of T cell (specific or non-specific) and toward a given type of T cells. For instance, an estimate λ1 = 0.58/h for the model in which only PyTCR cells recruit assumes that PyTCR cells recruit specific and non-specific T cells at the same rate. In another model notation “PyTCR:OT1” denotes the recruitment rate induced by PyTCR cells for OT1 cells. Numbers in parentheses indicate 95% confidence intervals for parameter estimates. Bold value indicates the weight of the best fit model.