Stability ranking

Share:

Description

Ranking variables based on their stability

Details

Detecting all relevant variables from a data set is challenging, especially when only few samples are available and data is noisy. Stability ranking provides improved variable rankings of increased robustness using resampling or subsampling.

Author(s)

juliane.siebourg@bsse.ethz.ch

Examples

 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
38
39
40
41
42
# generate dataset
d<-replicate(4,sample(1:10,10,replace=FALSE))
rownames(d)<-letters[1:10]

# rank aggregation on the dataset using two base methods
aggregRank(d, method='mean')
aggregRank(d, method='median')

# calculate summary statistic from the data
summaryStats(d, method='mean')
summaryStats(d, method='RSA')

# calculating replicate scores from different summary statistics
scores<-getSampleScores(d,'mean',decreasing=FALSE,bootstrap=TRUE)
scores<-getSampleScores(d,'mwtest',decreasing=FALSE,bootstrap=TRUE)

# perform RSA analysis

# get RSA format of data
rsaData<-dataFormatRSA(d)
# set RSA options
opts<-list(LB=min(d),UB=max(d),reverse=FALSE)
# run the RSA analysis
r<-runRSA(rsaData,opts)
# directly obtain the per gene RSA ranking from the data
r<-uniqueRSARanking(rsaData,opts)

# get stable Ranking, stable setsizes and the Pi matrix for default settings
# and stability threshold of 0.9
s<-getStability(d,0.9)

# run default stability ranking
s<-stabilityRanking(d)

# using an accessor function on the RankSummary object
stabRank(s)

# summarize a RankSummary object
summary(s)

# generate a rank matrix from a RankSummary object
getRankmatrix(s)