__________________________________________________________________________________________________________________________________________________________
Tellipsoid (version 08.1)
Paper. Keyur Desai,
J.R. Deller, Jr. and J. Justin McCormick. Tellipsoid:
Exploiting inter-gene correlation for improved
detection of differential gene expression, submitted to Bioinformatics.
preprint. [pdf]
What is it? Tellipsoid is a powerful statistical algorithm fordetecting differential gene
expression. The basic idea is to exploit inter-gene correlation
to share information across gene summary statistics, but while doing so
we also make use of identifiability-- the
fact that in most microarray data sets, a
large proportion of genes can be identified a priori as
non-differential on the bases of their t-statistics. We have just begun
to understand the full implications of this simple but very
different approach. Combining correlation with
identifiability is the key here.
What can it do? Current version supports two-sample
microarray studies. However, the core Tellipsoid methodology is much general.
SIMPLE-MAX Simultaneous Inference of Microarrays based on the PrincipLe of Entropy MAXimization.
Paper. Keyur Desai,
J.R. Deller, Jr. and J. Justin McCormick. THE DISTRIBUTION OF THE NUMBER OF FALSE DISCOVERIES IN HIGHLY CORRELATED DNA MICROARRAY DATA, submitted to Annals of Applied Statistics.
preprint. [pdf]
What is it? SIMPLEMAX is a method to ascertain the correct null distribution to tail counts of z values in the presence
of substantial inter-gene correlation. Our basic goal in the SIMPLEMAX
paper is to extend the second-moment theory for the null z-value
histogram in [Efron (2007) Correlation and Large-Scale Simultaneous Signifcance
Testing. Journal of the American Statistical Association,
102(477):93--103, 2007]to
include third-moment skewness corrections. Efron, in the same paper,
noticed the limitations of the 2nd-moment theory in some test cases;
this motivated the research behind SIMPLEMAX.
What can it do? Current version supports two-sample
microarray studies.
SIMPLEMAX takes a gene expression matrix, microarray labels, and a
tail-area and estimates an entire distribution of the number of false
discoveries falling in that tail-area.
MATLAB: simplemax.zip
(I have been tied up with Tellipsoid project for a while but in near
future I plan to provide a more user friedly version of simplemax.) __________________________________________________________________________________________________________________________________________________________
