Description Usage Arguments Details Value Author(s) References Examples
Test each transcript, gene, exon, or intron in a ballgown object for differential expression, using comparisons of linear models.
1 2 3 4 5  stattest(gown = NULL, gowntable = NULL, pData = NULL, mod = NULL,
mod0 = NULL, feature = c("gene", "exon", "intron", "transcript"),
meas = c("cov", "FPKM", "rcount", "ucount", "mrcount", "mcov"),
timecourse = FALSE, covariate = NULL, adjustvars = NULL, gexpr = NULL,
df = 4, getFC = FALSE, libadjust = NULL, log = TRUE)

gown 
name of an object of class 
gowntable 
matrix or matrixlike object with 
pData 
Required if 
mod 
object of class 
mod0 
object of class 
feature 
the type of genomic feature to be tested for differential
expression. If 
meas 
the expression measurement to use for statistical tests. Must be
one of 
timecourse 
if 
covariate 
string representing the name of the covariate of interest
for the differential expression tests. Must correspond to the name of a
column of 
adjustvars 
optional vector of strings representing the names of
potential confounders. Must correspond to names of columns of

gexpr 
optional data frame that is the result of calling

df 
degrees of freedom used for modeling expression over time with
natural cubic splines. Default 4. Only used if 
getFC 
if 
libadjust 
librarysize adjustment to use in linear models. By default,
the adjustment is defined as the sum of the sample's log expression
measurements below the 75th percentile of those measurements. To use
a different librarysize adjustment, provide a numeric vector of each
sample's adjustment value. Entries of this vector correspond to samples in
in rows of 
log 
if 
At minimum, you need to provide a ballgown object or count table, the type of feature you want to test (gene, transcript, exon, or intron), the expression measurement you want to use (FPKM, cov, rcount, etc.), and the covariate of interest, which must be the name of one of the columns of the 'pData' component of your ballgown object (or provided pData). This covariate is automatically converted to a factor during model fitting in nontimecourse experiments.
By default, models are fit using log2(meas + 1)
as the outcome for
each feature. To disable the log transformation, provide 'log = FALSE' as
an argument to 'stattest'. You can use the gowntable
option if you'd
like to to use a different transformation.
Library size adjustment is performed by default by using the sum of the log
nonzero expression measurements for each sample, up to the 75th percentile
of those measurements. This adjustment can be disabled by setting
libadjust=FALSE
. You can use mod
and mod0
to specify
alternative library size adjustments.
mod
and mod0
are optional arguments. If mod
is
specified, you must also specify mod0
. If neither is specified,
mod0
defaults to the design matrix for a model including only a
librarysize adjustment, and mod
defaults to the design matrix for a
model including a librarysize adjustment and covariate
. Note that
if you supply mod
and mod0
, covariate
,
timecourse
, adjustvars
, and df
are ignored, so make
sure your covariate of interest and all appropriate confounder
adjustments, including library size, are specified in mod
and
mod0
. By default, the librarysize adjustment is the sum of all
counts below the 75th percentile of nonzero counts, on the log scale
(log2 + 1).
Full model details are described in the supplement of http://biorxiv.org/content/early/2014/03/30/003665.
data frame containing the columns feature
, id
representing feature id, pval
representing the pvalue for testing
whether this feature was differentially expressed according to
covariate
, and qval
, the estimated false discovery rate
using this feature's signal strength as a significance cutoff. An
additional column, fc
, is included if getFC
is TRUE
.
Jeff Leek, Alyssa Frazee
http://biorxiv.org/content/early/2014/03/30/003665
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  data(bg)
# twogroup comparison:
stat_results = stattest(bg, feature='transcript', meas='FPKM',
covariate='group')
# timecourse test:
pData(bg) = data.frame(pData(bg), time=rep(1:10, 2)) #dummy time covariate
timecourse_results = stattest(bg, feature='transcript', meas='FPKM',
covariate='time', timecourse=TRUE)
# timecourse test, adjusting for group:
group_adj_timecourse_results = stattest(bg, feature='transcript',
meas='FPKM', covariate='time', timecourse=TRUE, adjustvars='group')
# custom model matrices:
### create example data:
set.seed(43)
sex = sample(c('M','F'), size=nrow(pData(bg)), replace=TRUE)
age = sample(21:52, size=nrow(pData(bg)), replace=TRUE)
### create design matrices:
mod = model.matrix(~ sex + age + pData(bg)$group + pData(bg)$time)
mod0 = model.matrix(~ pData(bg)$group + pData(bg)$time)
### build model:
adjusted_results = stattest(bg, feature='transcript', meas='FPKM',
mod0=mod0, mod=mod)

Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.