Contents
Phylogenetic and phenotypic variability
Estimates for missing parameters
Cross-validation details
Phylogenetic and phenotypic variability
PhyloPars first estimates the parameters of the evolutionary model, i.e.,
the phylogenetic covariances and phenotypic variances. These are subsequently used to estimate missing feature values and
to perform cross-validation.
feature |
phylogenetic s.d. |
phenotypic s.d. |
% variance explained by phylogeny |
cross-validation results |
|
error |
bias |
maximum growth rate | 0.2915 | ln /d | 0.2853 | ln /d | 85 | % | 0.344 | ln /d | -0.00227 | ln /d |
diameter | 0.2638 | ln µm | 0.3586 | ln µm | 65 | % | 0.366 | ln µm | -0.0125 | ln µm |
surface area | 0.6726 | ln µm² | 1.094e-05 | ln µm² | 100 | % | 0.22 | ln µm² | 0.00592 | ln µm² |
phosphate affinity | 0.9468 | ln /d/µM | 0.9311 | ln /d/µM | 78 | % | 1.08 | ln /d/µM | -0.053 | ln /d/µM |
volume | 0.913 | ln µm³ | 0.588 | ln µm³ | 92 | % | 0.797 | ln µm³ | -0.0336 | ln µm³ |
length | 0.4895 | ln µm | 0.2509 | ln µm | 95 | % | 0.455 | ln µm | -0.0436 | ln µm |
Phylogenetic correlations
| maximum growth rate | diameter | surface area | phosphate affinity | volume | length |
maximum growth rate | 1 | | | | | |
diameter | -0.412 | 1 | | | | |
surface area | -0.558 | 0.763 | 1 | | | |
phosphate affinity | 0.193 | 0.516 | 0.034 | 1 | | |
volume | -0.604 | 0.742 | 0.979 | -0.105 | 1 | |
length | -0.186 | 0.745 | 0.686 | 0.700 | 0.539 | 1 |
Phylogenetic regression coefficients
Rows represent independent variables and columns dependent variables. A regression of feature i on j thus can be found at row j, column i.
| maximum growth rate | diameter | surface area | phosphate affinity | volume | length |
maximum growth rate | 1 | -0.372 | -1.286 | 0.628 | -1.893 | -0.312 |
diameter | -0.455 | 1 | 1.946 | 1.854 | 2.567 | 1.382 |
surface area | -0.242 | 0.299 | 1 | 0.048 | 1.329 | 0.500 |
phosphate affinity | 0.060 | 0.144 | 0.024 | 1 | -0.102 | 0.362 |
volume | -0.193 | 0.214 | 0.721 | -0.109 | 1 | 0.289 |
length | -0.111 | 0.401 | 0.943 | 1.354 | 1.005 | 1 |
Estimates for missing parameters
Using the optimal phylogenetic covariances and phenotypic variances listed above, PhyloPars estimates the values
that were originally missing in the feature matrix. The table below lists all estimated
values. If one or more observations have been provided for a value, this is denoted by a trailing
asterisk symbol (*). You can click on an entry to view the contribution of individual observations to
the estimate, and to retrieve details such as the standard deviation of
the estimate. You can also download the complete table as a single text file.
Cross-validation details
In addition to estimating missing values, PhyloPars performs cross-validation:
it temporarily excludes each observation from the available information, in order to
calculate its best estimate given all other observations and the optimal phylogenetic covariances and phenotypic variances.
This provides a detailed estimate of the error and bias one may expect for the
estimated missing feature values. The distributions of estimation errors as calculated
through cross-validation are shown below; these are compared with two null models (red and green
curves) in order to assess the improvement resulting from the PhyloPars evolutionary model.
Maximum growth rate
Cross-validation error
|
| evolutionary model | mean model | nearest neighbor model |
mean bias | -0.00227 | ln /d | 8.81e-18 | ln /d | 0.0277 | ln /d |
mean error | 0.344 | ln /d | 0.583 | ln /d | 0.459 | ln /d |
Diameter
Cross-validation error
|
| evolutionary model | mean model | nearest neighbor model |
mean bias | -0.0125 | ln µm | 4.31e-16 | ln µm | -0.205 | ln µm |
mean error | 0.366 | ln µm | 0.495 | ln µm | 0.622 | ln µm |
Surface area
Cross-validation error
|
| evolutionary model | mean model | nearest neighbor model |
mean bias | 0.00592 | ln µm² | -2.96e-16 | ln µm² | -0.285 | ln µm² |
mean error | 0.22 | ln µm² | 1.62 | ln µm² | 1.45 | ln µm² |
Phosphate affinity
Cross-validation error
|
| evolutionary model | mean model | nearest neighbor model |
mean bias | -0.053 | ln /d/µM | 8.73e-16 | ln /d/µM | 0.00903 | ln /d/µM |
mean error | 1.08 | ln /d/µM | 1.5 | ln /d/µM | 1.29 | ln /d/µM |
Volume
Cross-validation error
|
| evolutionary model | mean model | nearest neighbor model |
mean bias | -0.0336 | ln µm³ | -4.63e-16 | ln µm³ | -0.0518 | ln µm³ |
mean error | 0.797 | ln µm³ | 1.76 | ln µm³ | 1.17 | ln µm³ |
Length
Cross-validation error
|
| evolutionary model | mean model | nearest neighbor model |
mean bias | -0.0436 | ln µm | 7.98e-16 | ln µm | -0.0568 | ln µm |
mean error | 0.455 | ln µm | 0.951 | ln µm | 0.776 | ln µm |