Description Usage Arguments Details Value Author(s) References See Also Examples
View source: R/hrunbiasedDiagnostic.r
Contains several diagnostic functions for adjusted Cox proportional hazard models using gene signatures
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 | hrunbiasedDiagnostic(nda, correction.type = "NCsignatures",
score, signature.method = "zscore",
seqquant = c(0,seq(0.05,1,length.out=6)), sigSize = 50,
diagnostic.type = c("density.rs", "permutations", "simulations",
"positive.cont.power", "PCAcorrelation"), ...)
hrunbiasedDensityRS(nda, correction.type = "NCsignatures", score,
seqquant = c(0,seq(0.05,1,length.out=6)), sigSize = 50,
nrs = 5000, comp.RUV = c(1,2), adj.var.GS = "GSadj",
adjusted.var.random = NA, adjusted.var.fixed= NA,
executation.info = TRUE, mc.cores = 1, GSs = NULL, rg.cox = TRUE,
id.size = 1, basic.cont = TRUE, signature.method = "zscore", ...)
hrunbiasedPermutations(nda, correction.type = "NCsignatures",
score, seqquant = c(0,seq(0.05,1,length.out=6)), sigSize = 50,
nrs = 5000, comp.RUV = c(1,2), nperm = 100,
adj.var.GS = "GSadj", permutation.type = "none", GSassociation = TRUE,
executation.info = TRUE, mc.cores = 1, GSs = NULL, basic.cont = TRUE,
GS = NULL, signature.method = "zscore", ...)
hrunbiasedPCpower(nda, positive.controls, correction.type = "NCsignatures", score,
seqquant = c(0,seq(0.05,1,length.out=6)), alternative = "two.sided",
bootstrapping = TRUE, prop.pos.cont = 1, sigSize = 50,
id.size = 1, nrs = 5000, nrpcs = 500, comp.RUV = c(1,2),
null.dist = "asymptotic", adj.var.GS = "GSadj", adjusted.var.random = NA,
adjusted.var.fixed= NA, executation.info = TRUE, mc.cores = 1, GSs = NULL,
cdRS = NULL, basic.cont = TRUE, ...)
hrunbiasedSimulations(nda, positive.controls, GScoef = 0,
tscoef = 0, ninst.rs = 10, ninst.ts = 500, nrs = 500, lambdas = c(0.1,0.2),
gammas = c(1.5,2), mc.cores = 1, stat = TRUE,
executation.info = TRUE, GSs = NULL, signature.method = "zscore",
basic.cont = TRUE, sigSize = 50, correction.type = "none", ...)
hrunbiasedPCAcorrelation(nda, nrs = 100, sigSize = 50, mc.cores = 1,
comp.PCA = c(1,2), GS = NULL, executation.info = TRUE,
signature.method = "zscore", ...)
|
nda |
|
correction.type |
Standard Cox model can be used when
|
score |
Vector of the same length as number of rows in |
signature.method |
Signature score method: to choose from
|
seqquant |
Numeric vector with values between zero and one used to set negative controls |
diagnostic.type |
Diagnostic to be performed: at least one among
|
sigSize |
Numeric vector of at least one element with signature
sizes to be considered (a single signature size is used for
|
nrs |
Number of random signature instances used for approximated null distribution |
comp.RUV |
Factors used in factor analysis signature correction
when |
GSs |
List with correction variables. If NULL, GSs is found by
the approach specified in |
adj.var.GS |
Only if correction.type is not "none" and siganture.method is "zscore". Transforming signature scores by subtracting the correction variable scores "GScor", residuals of linear model using the correction variable as covariate "GSlmcor", or using the correction variable in the Cox model "GSadj" (default) |
adjusted.var.random |
Variables that enter in the Cox model as random effect |
adjusted.var.fixed |
Confounding variables that enter in the Cox model as fixed effect |
executation.info |
If |
mc.cores |
The number of cores to use for parallel executions |
rg.cox |
If TRUE, survival effects at gene level are computed |
alternative |
Character string specifying the alternative
hypothesis, must be one of |
nperm |
Number of outcome shuffle instances to be done |
permutation.type |
Either |
GSassociation |
GS association with survival is computed to show global tendency |
GS |
List with the global signature (for internal use) |
positive.controls |
Character string with gene signature to be
used as positive control (must match |
bootstrapping |
If |
prop.pos.cont |
Numeric vector of at least one element that indicates the amount (proportion between 0 and 1) of purity of the positive control signatures to be tested (one by default, zero would correspond to random signatures, and in betweeen there would be contaminated positive control signatures) |
nrpcs |
Number of random positive control signature instances to be used to approximate power |
id.size |
Signature size used |
null.dist |
Either |
cdRS |
A |
basic.cont |
Basic controls are performed, these are done no
matter the |
GScoef |
The log HR effect of global signature (common tendency) for simulations |
tscoef |
The log HR effect of target signature (unique tendency) for simulations |
ninst.rs |
Instances of random signature distributions for simulated data analysis (must be larger than zero) |
ninst.ts |
Instances of target signature estimated effect for simulated data analysis (can be zero) |
lambdas |
Weibull distribution shape parameters (see
|
gammas |
Weibull distribution scale parameters (see
|
stat |
If |
comp.PCA |
Numeric(2) vector with the specific components to compute correlation between random signatures and the projections |
... |
Arguments passed to or from other methods to the low level. |
Several diagnostic plots via plot.hrunbiasedDiagnostic
are available which can be useful to decide whether an
adjustment in estimating hazard ratios for gene signatures is
needed and, in some extent, which class of correction could be
appropriate. The diagnostics implemented in this function can be
grouped in five modules: (i) density random signatures; (ii)
permutated outcome; (iii) positive control power; (iv)
simulated survival data; and (v) random signatures correlation
with PC. Detailed information about the plots in each of the
modules is provided in plot.hrunbiasedDiagnostic
.
Correction types include adjustments by global signature
("Gsignature"
), negative control signatures
("NCsignatures"
), or negative control factors ("RUV"
).
Global signature is a summary across all standardized genes in the
data whereas negative control adjustments are found by summarising the
information of only a few particular genes. All genes are given a
score via attribute score
(i.e., gene variability, cross-study
reproducibility,...).
Selected negative control genes are those whose score is smaller than
the percentile indicated in seqquant
. Several
values for seqquant
can be given so the diagnostic can be done
over different specifications of negative control genes.
Permutations based diagnostic is done to show global biases in random
signature distributions due to both sampling and gene-to-gene
dependence structure.
The permutation.type
equal to "none" or "Gsignature" have the
same interpretation as the one for the correction type. Besides, in
"norm01" we also consider a shuffling of the samples independently
for all genes in the data so gene-to-gene relationships are broken.
If diagnostic.type = "density.rs"
or hrunbiasedDensityRS function
is called, return an object of class hrunbiasedDensityRS
with
attributes
"hr.rs" |
Cox model outcome of random signatures |
"geneMeanHR.signatureHR" |
Average log HR genewise in random signature and log HR in the same random signatures |
If diagnostic.type = "permutations"
or hrunbiasedPermutations function
is called, return an object of class hrunbiasedPermutations
with
attributes
"perms" |
Time-to-event data permuted indexes |
"perm.out" |
Outcome of permutations study |
If diagnostic.type = "positive.cont.power"
or hrunbiasedPCpower
function is called, return an object of class hrunbiasedPCpower
with
attributes
"hr.pc" |
Cox model outcome of positive control signature |
"pvals.pc" |
P-values found for positive control signatures |
If diagnostic.type = "simulations"
or hrunbiasedSimulations function
is called, return an object of class hrunbiasedSimulations
with
attributes
"simu.info" |
Simulated data analysis output for diagnostic plot |
If diagnostic.type = "PCAcorrelations"
or hrunbiasedPCAcorrelation
function is called, return an object of class hrunbiasedPCAcorrelation
with attributes
"rs.cor" |
Correlation between random signatures and the projections of the comp.PCA components |
"rs.gs.cor" |
Correlation between random signatures adjusted by GS and the projections of the comp.PCA components |
"gs.cor" |
Correlation between Global signature and the projections of the comp.PCA components |
For all called functions or diagnostic.type
options, the
following items are also included
"GS" |
Global signature |
"seqquant" |
Sequence seqquant used |
"sigSize" |
Signature size |
"correction.type" |
Correction approach used |
"diagnostic.type" |
Diagnostic type used |
Adria Caballe Mestres
Caballe Mestres A, Berenguer Llergo A and Stephan-Otto Attolini C. Adjusting for systematic technical biases in risk assessment of gene signatures in transcriptomic cancer cohorts. bioRxiv (2018).
Gagnon-Bartsch J. and Speed T. Using control genes to correct for unwanted variation in microarray data. Biostatistics 13, 539-552 (2012).
Hanzelmann S., Castelo R. and Guinney J. GSVA: gene set variation analysis for microarray and RNA-Seq data. BMC Bioinformatics 7 (2013).
plot.hrunbiasedDiagnostic
and
hrunbiasedTesting
1 2 3 4 5 6 7 | eh <- ExperimentHub()
nda.brca <- query(eh, "mcsurvdata")[["EH1497"]]
nda.gse1456 <- nda.brca[,nda.brca$dataset=="GSE1456"]
out.brca.GSE1456 <- hrunbiasedDiagnostic(nda.gse1456,
correction.type = "none", mc.cores = 1, diagnostic.type = "density.rs",
sigSize = 50, nrs = 100, id.size = 1, executation.info = FALSE)
|
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.