Function for computing significance of clustering
$p$-value. $p$-value is obtained from
sigclust, a simulation based procedure for
testing significance of clustering in high dimension low
sample size (HDLSS) data.
The SigClust hypothesis test is given:
H0: data generated from single Gaussian
H1: data
not generated from single Gaussian
Usage
SFpval(data, normalize = 1, flag = 1)
Arguments
data
a $d x n$ matrix of read
counts at $d$ positions for $n$ samples.
normalize
a $n x 1$ logical vector
of flagged samples.
flag
a $n x 1$ logical vector of
samples flagged as low expression. If flag == 1,
default low expression cutoffs are applied to
data. If flag == 0, no samples are flagged
as low expression (equivalent to setting flag =
zeros(n,1)).
Value
SFpval returns an object of class
sigclust-class. Avaliable slots are
described in detail in the sigclust
package. Primarily, we make use of @pvalnorm.