MOESM2 of Quantitation of the latent HIV-1 reservoir from the sequence diversity in viral outgrowth assays

Additional file 2. Fig. S2. Effect of misspecified prior distributions on posterior estimates of IUPM from real data sets. The filled curves represent the same posterior distributions as in Fig. 5, where the prior distributions on variant frequencies were set to α = {10, 1,… , 1} for patient 106 and α = {1,…, 1} for patient 111. The dashed curves represent the posterior distributions obtained when these priors were swapped between the patient data sets. These results illustrate that misspecification of the prior distribution on variant frequencies can have a measurable effect on posterior estimates of IUPM where the underlying variant frequencies are skewed toward a single common variant (patient 106). However when the virus population has low clonality and there is a mixture of positive and negative wells at the lowest dilution of the QVOA (patient 111, see Table 2), the posterior estimates are more robust to the prior settings