TABLE 3

Fitting fluctuation data to a two-parameter model of postplating growth

α-Factor resistance10× canavanine resistance5-FOA resistance
CloneMut. rate (10−6)Div. postplatingP(LLR)AICMut. rate (10−7)Div. postplatingP(LLR)AICMut. rate (10−8)Div. postplating
A3.071.420.0005−10.01.760.370.260.76.490
B3.660.950.01−4.11.161.080.005−6.04.770
C4.170.960.01−4.71.50.940.02−3.77.190
D2.892.45<0.0001−14.41.191.120.007−5.24.970.03
E2.142.74<0.0001−24.61.530.670.03−2.74.480
F3.341.480.0002−11.61.460.540.1−0.66.70
G2.941.460.0007−9.41.690.080.741.94.740
H2.612.32<0.0001−20.41.880.180.491.55.010
I2.562.45<0.0001−19.81.480.420.210.47.030
J3.361.170.001−8.11.550.720.04−2.23.050
Average ± SD3.07 ± 0.591.74 ± 0.681.52 ± 0.230.61 ± 0.365.44 ± 1.34<0.01
Combined3.071.66<0.0001−135.31.520.57<0.0001−26.85.430
  • Mut. rate is the phenotypic mutation per genome per generation. Div. postplating is the total number of cell divisions after plating that can give rise to mutants. We show two statistical tests for whether the fit is improved by using a two-parameter model that includes the generation of mutants after plating; P(LLR) is the probability from a log-likelihood-ratio test that the one-parameter model should be preferred, and AIC reports the values of the Akaike information criterion, where negative values indicate that the two-parameter model is preferred. Italic values indicate a preference for the two-parameter model. Both tests are explained in more detail in materials and methods. The combined data set treats the 10 72-tube fluctuation assays as one 720-tube fluctuation assay.