Comments (2)
original patternMarkers version for inspiration
#' patternMarkers
#'
#' @param Amatrix A matrix of genes by weights resulting from CoGAPS or other NMF decomposition
#' @param scaledPmatrix logical indicating whether the corresponding pattern matrix was fixed to have max 1 during decomposition
#' @param Pmatrix the corresponding Pmatrix (patterns X samples) for the provided Amatrix (genes x patterns). This must be supplied if scaledPmatrix is FALSE.
#' @param threshold the type of threshold to be used. The default "cut" will thresholding by the first gene to have a lower ranking, i.e. better fit to, a pattern. Alternatively, threshold="all" will return all of the genes in rank order for each pattern.
#' @param lp a vector of weights for each pattern to be used for finding markers. If NA markers for each pattern of the A matrix will be used.
#' @param full logical indicating whether to return the ranks of each gene for each pattern
#'
#' @return By default a non-overlapping list of genes associated with each \code{lp}. If \code{full=TRUE} a data.frame of
#' genes rankings with a column for each \code{lp} will also be returned.
#' @export
#'
#' @examples \dontrun{
#' patternMarkers(Amatrix=AP$Amean,scaledPmatrix=FALSE,Pmatrix=NA,threshold="cut")
#' }
#'
patternMarkers <- function(
Amatrix=NA, #A matrix of genes by weights resulting from CoGAPS or other NMF decomposition
scaledPmatrix=FALSE, # logical indicating whether the corresponding pattern matrix was fixed to have max 1 during decomposition
Pmatrix=NA, #the corresponding Pmatrix (patterns X samples) for the provided Amatrix (genes x patterns). This must be supplied if scaledPmatrix is FALSE.
threshold="cut", # the type of threshold to be used. The default "cut" will thresholding by the first gene to have a lower ranking, i.e. better fit to, a pattern. Alternatively, threshold="all" will return all of the genes in rank order for each pattern.
lp=NA, # a vector of weights for each pattern to be used for finding markers. If NA markers for each pattern of the A matrix will be used.
full=FALSE #logical indicating whether to return the ranks of each gene for each pattern.
){
if(scaledPmatrix==FALSE){
if(!is.na(Pmatrix)){
pscale <- apply(Pmatrix,1,max) # rescale p's to have max 1
Amatrix <- sweep(Amatrix, 2, pscale, FUN="*") # rescale A in accordance with p's having max 1
}
else(warning("P values must be provided if not already scaled"))
}
# find the A with the highest magnitude
Arowmax <- t(apply(Amatrix, 1, function(x) x/max(x)))
pmax<-apply(Amatrix, 1, max)
# determine which genes are most associated with each pattern
ssranks<-matrix(NA, nrow=nrow(Amatrix), ncol=ncol(Amatrix),dimnames=dimnames(Amatrix))#list()
ssgenes<-matrix(NA, nrow=nrow(Amatrix), ncol=ncol(Amatrix),dimnames=NULL)
nP=dim(Amatrix)[2]
if(!is.na(lp)){
if(length(lp)!=dim(Amatrix)[2]){
warning("lp length must equal the number of columns of the Amatrix")
}
sstat <- apply(Arowmax, 1, function(x) sqrt(t(x-lp)%*%(x-lp)))
ssranks[order(sstat),i] <- 1:length(sstat)
ssgenes[,i]<-names(sort(sstat,decreasing=FALSE))
} else {for(i in 1:nP){
lp <- rep(0,dim(Amatrix)[2])
lp[i] <- 1
sstat <- apply(Arowmax, 1, function(x) sqrt(t(x-lp)%*%(x-lp)))
ssranks[order(sstat),i] <- 1:length(sstat)
ssgenes[,i]<-names(sort(sstat,decreasing=FALSE))
}}
if(threshold=="cut"){
pIndx<-apply(ssranks,1,which.min)
ssgenes.th <- lapply(unique(pIndx),function(x) names(pIndx[pIndx==x]))
}
if(threshold=="All"){
geneThresh <- apply(sweep(ssranks,1,t(apply(ssranks, 1, min)),"-"),2,function(x) which(x==0))
ssgenes.th <- lapply(geneThresh,names)
}
if(full){return(list("PatternMarkers"=ssgenes.th,"PatternRanks"=ssranks))
} else{return("PatternMarkers"=ssgenes.th)}
}
from cogaps.
the results between the current (#102) patternMarkers and the original has diverted:
data(GIST)
res <- CoGAPS(GIST.data_frame, nIterations=100, seed=1, messages=FALSE)
cut:
> pm_current_cut <- patternMarkers(res, threshold = "cut", axis = 1, lp = NA)
> pm_orig_cut <- patternMarkers.orig(Amatrix=res@featureLoadings, Pmatrix=t(res@sampleFactors), threshold = "cut", lp = NULL)
> lengths(pm_current_cut$PatternMarkers)
Pattern_1 Pattern_2 Pattern_3 Pattern_4 Pattern_5 Pattern_6 Pattern_7
14 46 10 270 27 6 37
> lengths(pm_orig_cut)
[1] 183 326 82 237 145 185 205
all:
> pm_current_all <- patternMarkers(res, threshold = "all", axis = 1, lp = NA)
> pm_orig_all <- patternMarkers.orig(Amatrix=res@featureLoadings, Pmatrix=t(res@sampleFactors), threshold = "All", lp = NULL)
> lengths(pm_current_all$PatternMarkers)
Pattern_1 Pattern_2 Pattern_3 Pattern_4 Pattern_5 Pattern_6 Pattern_7
21 6 1 1047 57 6 225
> lengths(pm_orig_all)
Pattern_1 Pattern_2 Pattern_3 Pattern_4 Pattern_5 Pattern_6 Pattern_7
183 205 185 327 149 240 82
from cogaps.
Related Issues (20)
- CoGAPS does not learn to specifed nPatterns when runnign in dsitributed mode HOT 4
- Enable the Checkpoints HOT 2
- patternMarkers 'all' returning empty lists HOT 4
- Refactor - add comments
- getPatternGeneSet gives wrong gene.set names HOT 4
- fix `threshold="cut"` mode
- Warning: Large values detected, is data log transformed? HOT 1
- Seeking Clarification on Functions in the CoGAPS Package HOT 2
- Confusion Regarding patternMarkers Results HOT 2
- Cpp tests
- Standard deviation when running distributed
- Vignettes fail on querying ENSEMBL
- Error when performing getPatternHallmarks: Error in 'collect()' HOT 9
- Replace Ensemble dependency and Add overrepresentation and enrichment tests for gene sets in patterns
- Confusion on the parallelization of R CoGAPS HOT 5
- GSEA algorithms for the function getPatternHallmarks HOT 9
- default number of patterns
- Too long runtime for GoGAPS R HOT 2
- Semisupervised CoGAPS
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from cogaps.