NEWS
st 1.2.7 (2021-11-27)
st 1.2.6 (2020-09-24)
- change modt.fun() to use limma::eBayes() rather than limma::ebayes().
st 1.2.5 (2015-07-08)
- added import statements required by R-devel.
st 1.2.4 (2015-03-21)
- fixed the "please use :: or requireNamespace() instead" warnings.
st 1.2.3 (2014-11-14)
- change of maintainer email address.
- fixed the "no visible global function definition" warnings.
st 1.2.2 (2014-04-18)
- added import statement in NAMESPACE to address corresponding warnings
in R 3.1.0.
st 1.2.1 (2013-09-01)
- package update necessary due to updated functions in sda version 1.3.1.
In particular, sda::centroids() and sda::catscore() functions now
use shrinkage estimates of frequencies.
Correspondingly, diffmean.fun(), studentt.fun(), samL1.R(), lait.R()
efront.R() and cst.R() have been modified internally to continue to
use empirical frequencies. In addition, shrinkt.fun(), shrinkt.stat()
and shrinkcat.stat() now have a new "lambda.freqs" option.
st 1.2.0 (2012-12-04)
- one-sample shrinkage CAT scores and paired shrinkage CAT scores are now
implemented, as well as CAT scores with unequal variances. In addition,
the regularization parameters lambda and lambda.var can now be specified.
- similary, estimation of one-sample shrinkage t-scores and paired
shrinkage t-scores is now available, as well as optional specification
of lambda.var regularization parameter.
- in addtion, computation of the paired student t statistics as well as
student t scores with unequal variance is now possible.
st 1.1.8 (2012-08-20)
- changes to work with the updated centroids() and catscore()
functions in sda version 1.2.2
st 1.1.7 (2012-01-22)
st 1.1.6 (2011-07-05)
- small change in studentt.fun and diffmean.fun to avoid uncessary
verbose output
st 1.1.5 (2011-06-28)
- small changes in the code needed to work with the modified centroids
function from the sda package version 1.2.0
- neighborhood options removed from shrinkcat.stat()
- shrinkcat.stat() now uses catscore() function from sda package
(this is much faster than the version previously implemented in "st")
st 1.1.4 (2010-05-06)
- added option "group.method". Now groups are determined either
by the correlation neighborhood (possibly overlapping groups) or
by the clusters on the correlation graph (disjoint groups).
st 1.1.3 (2009-05-10)
- added option "group.thresh" to allow the specification of
correlation neighborhoods around features and the computation
of the grouped cat score.
st 1.1.2 (2009-04-26)
- correlation-shared t-statistic of Tibshirani and Wasserman
(2006) added: new functions cst.stat() and cst.fun().
- correlation-predicted t-statistic of Lai (2008) added:
new functions lait.stat(), lait.fun() and lai.tscore().
st 1.1.1 (2009-02-05)
- rankprod.stat() and rankprod.fun() added for calculation of the
two-sided rank products statistic.
- reference to Zuber and Strimmer (2009) added.
- small corrections in the help pages, to pass the more stringent
checks on .Rd files introduced in R in January 2009.
st 1.1.0 (2008-12-13)
- new functions shrinkcat.stat() und shrinkcat.fun() implementing
a shrinkage estimate of the "correlation-adjusted t-score" described
in Zuber and Strimmer (2009).
- part of the code was rewritten to share code with the "sda" package
(exploiting the link between discriminant analysis and gene ranking).
st 1.0.3 (2008-10-27)
- the class labels can now be specified as arbitrary factors,
the argument L is not any more restricted to be a vector containing
only 1s and 2s.
st 1.0.2 (2007-10-21)
- some corrections and extensions have been made in the documentation,
in particular, how to compute q-values and local FDR values
- shrinkt() now returns a proper vector (one for which is.vector() is TRUE).
- the license was changed from "GNU GPL 2 or later" to "GNU GPL 3 or later".
st 1.0.1 (2007-02-28)
- reference to Opgen-Rhein and Strimmer (2007) updated.
- updated email address
- updated minimum R versions and version of corpcor package
st 1.0.0 (2006-08-31)
This is the first standalone release (31 August 2006
This package implements the "shrinkage t" statistic described
in Opgen-Rhein and Strimmer (2007). It also offers a convenient
interface to a number of other regularized t-type statistics often
used in high-dimensional case-control studies.