Description Usage Arguments Details Value Author(s) References See Also Examples
Performs t-tests to test for differences in the mean loadings between two sets of signatures found using GSLCCA. Corrects for multiple comparisons.
1 |
x, y |
a matrix of signatures from a GSLCCA analysis, with one column per subject. |
normalise |
logical value indicating whether or not to normalise the signatures first, so that the sum of squares is equal to one. |
p.adjust.method |
the method of adjustment for multiple comparisons, see
|
... |
further arguments passed to |
In GSLCCA, a signature is the set of loadings or coefficients, \mathbf{a}, of the multivariate response \mathbf{Y}, that give the highest correlation between the y scores \mathbf{Ya} and the fitted nonlinear model.
In the context of EEG analysis the signature represents the relative importance of each frequency in relation to the PK/PD model.
In a GSLCCA analysis with multiple subjects, a signature will have been estimated for each subject. This function compares the mean loading of each variable between two such sets of signatures, using t-tests.
When normalise=TRUE
, each signature is first scaled so that the sum of
squared loadings is equal to one.
The p-values from the t-test are corrected for multiplicity, using the method
given by p.adjust.method
, which specifies the false
discovery rate by default (Benjamini & Hochberg, 1995).
A list with the following components
call |
the call to |
x |
the first set of (normalised) signatures. |
y |
the second set of (normalised) signatures. |
pvalue |
the corrected pvalue. |
Heather Turner
Benjamini, Y., and Hochberg, Y. (1995). Controlling the false discovery rate: a practical and powerful approach to multiple testing. Journal of the Royal Statistical Society Series B, 57, 289–300.
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 | data(clonidine)
### Arbitrarily split the rats into two groups of four
### - expect results to be roughly the same!
grp1 <- subset(clonidine, Rat < 39)
grp2 <- subset(clonidine, Rat >= 39)
## fit same model to each group
result <- gslcca(spectra, "Double Exponential",
time = Time, subject = Rat, treatment = Treatment,
subject.smooth = TRUE, pct.explained = 0.96, data = grp1)
result2 <- gslcca(spectra, "Double Exponential",
time = Time, subject = Rat, treatment = Treatment,
subject.smooth = TRUE, pct.explained = 0.96, data = grp2)
## run snapshot analysis and plot
## - observed differences far from significant as expected
snap <- snapshot(signatures(result), signatures(result2))
par(mfrow = c(2,2))
plot(snap, type = c("signatures", "means", "pvalue"),
names = c("Group 1", "Group 2"))
### Artificially create a difference by swapping the data
### from the delta and theta bands for Group 1
grp1$spectra <- grp1$spectra[, c(5:8, 1:4, 9:36)]
colnames(grp1$spectra) <- colnames(grp2$spectra)
## refit model for group 1 and re-run snapshot analysis
result <- gslcca(spectra, "Double Exponential",
time = Time, subject = Rat, treatment = Treatment,
subject.smooth = TRUE, pct.explained = 0.96, data = grp1)
snap <- snapshot(signatures(result), signatures(result2))
## now shows highly significant difference in delta,
## significant difference in theta
## (Group 1 signatures less consistent here)
par(mfrow = c(2,2))
plot(snap, type = c("signatures", "means", "pvalue"),
names = c("Group 1", "Group 2"))
## compact display
par(mfrow = c(1,1))
plot(snap, type = "compact", names = c("Group 1", "Group 2"))
|
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.