comparison scripts/RaceID_class.R @ 0:e01c989c7543 draft default tip

planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/raceid commit 39918bfdb08f06862ca395ce58a6f5e4f6dd1a5e
author iuc
date Sat, 03 Mar 2018 17:34:16 -0500
parents
children
comparison
equal deleted inserted replaced
-1:000000000000 0:e01c989c7543
1 ## load required packages.
2 require(tsne)
3 require(pheatmap)
4 require(MASS)
5 require(cluster)
6 require(mclust)
7 require(flexmix)
8 require(lattice)
9 require(fpc)
10 require(amap)
11 require(RColorBrewer)
12 require(locfit)
13
14 ## class definition
15 SCseq <- setClass("SCseq", slots = c(expdata = "data.frame", ndata = "data.frame", fdata = "data.frame", distances = "matrix", tsne = "data.frame", kmeans = "list", background = "list", out = "list", cpart = "vector", fcol = "vector", filterpar = "list", clusterpar = "list", outlierpar ="list" ))
16
17 setValidity("SCseq",
18 function(object) {
19 msg <- NULL
20 if ( ! is.data.frame(object@expdata) ){
21 msg <- c(msg, "input data must be data.frame")
22 }else if ( nrow(object@expdata) < 2 ){
23 msg <- c(msg, "input data must have more than one row")
24 }else if ( ncol(object@expdata) < 2 ){
25 msg <- c(msg, "input data must have more than one column")
26 }else if (sum( apply( is.na(object@expdata),1,sum ) ) > 0 ){
27 msg <- c(msg, "NAs are not allowed in input data")
28 }else if (sum( apply( object@expdata,1,min ) ) < 0 ){
29 msg <- c(msg, "negative values are not allowed in input data")
30 }
31 if (is.null(msg)) TRUE
32 else msg
33 }
34 )
35
36 setMethod("initialize",
37 signature = "SCseq",
38 definition = function(.Object, expdata ){
39 .Object@expdata <- expdata
40 .Object@ndata <- expdata
41 .Object@fdata <- expdata
42 validObject(.Object)
43 return(.Object)
44 }
45 )
46
47 setGeneric("filterdata", function(object, mintotal=3000, minexpr=5, minnumber=1, maxexpr=500, downsample=FALSE, dsn=1, rseed=17000) standardGeneric("filterdata"))
48
49 setMethod("filterdata",
50 signature = "SCseq",
51 definition = function(object,mintotal,minexpr,minnumber,maxexpr,downsample,dsn,rseed) {
52 if ( ! is.numeric(mintotal) ) stop( "mintotal has to be a positive number" ) else if ( mintotal <= 0 ) stop( "mintotal has to be a positive number" )
53 if ( ! is.numeric(minexpr) ) stop( "minexpr has to be a non-negative number" ) else if ( minexpr < 0 ) stop( "minexpr has to be a non-negative number" )
54 if ( ! is.numeric(minnumber) ) stop( "minnumber has to be a non-negative integer number" ) else if ( round(minnumber) != minnumber | minnumber < 0 ) stop( "minnumber has to be a non-negative integer number" )
55 if ( ! ( is.numeric(downsample) | is.logical(downsample) ) ) stop( "downsample has to be logical (TRUE/FALSE)" )
56 if ( ! is.numeric(dsn) ) stop( "dsn has to be a positive integer number" ) else if ( round(dsn) != dsn | dsn <= 0 ) stop( "dsn has to be a positive integer number" )
57 object@filterpar <- list(mintotal=mintotal, minexpr=minexpr, minnumber=minnumber, maxexpr=maxexpr, downsample=downsample, dsn=dsn)
58 object@ndata <- object@expdata[,apply(object@expdata,2,sum,na.rm=TRUE) >= mintotal]
59 if ( downsample ){
60 set.seed(rseed)
61 object@ndata <- downsample(object@expdata,n=mintotal,dsn=dsn)
62 }else{
63 x <- object@ndata
64 object@ndata <- as.data.frame( t(t(x)/apply(x,2,sum))*median(apply(x,2,sum,na.rm=TRUE)) + .1 )
65 }
66 x <- object@ndata
67 object@fdata <- x[apply(x>=minexpr,1,sum,na.rm=TRUE) >= minnumber,]
68 x <- object@fdata
69 object@fdata <- x[apply(x,1,max,na.rm=TRUE) < maxexpr,]
70 return(object)
71 }
72 )
73
74
75 downsample <- function(x,n,dsn){
76 x <- round( x[,apply(x,2,sum,na.rm=TRUE) >= n], 0)
77 nn <- min( apply(x,2,sum) )
78 for ( j in 1:dsn ){
79 z <- data.frame(GENEID=rownames(x))
80 rownames(z) <- rownames(x)
81 initv <- rep(0,nrow(z))
82 for ( i in 1:dim(x)[2] ){
83 y <- aggregate(rep(1,nn),list(sample(rep(rownames(x),x[,i]),nn)),sum)
84 na <- names(x)[i]
85 names(y) <- c("GENEID",na)
86 rownames(y) <- y$GENEID
87 z[,na] <- initv
88 k <- intersect(rownames(z),y$GENEID)
89 z[k,na] <- y[k,na]
90 z[is.na(z[,na]),na] <- 0
91 }
92 rownames(z) <- as.vector(z$GENEID)
93 ds <- if ( j == 1 ) z[,-1] else ds + z[,-1]
94 }
95 ds <- ds/dsn + .1
96 return(ds)
97 }
98
99 dist.gen <- function(x,method="euclidean", ...) if ( method %in% c("spearman","pearson","kendall") ) as.dist( 1 - cor(t(x),method=method,...) ) else dist(x,method=method,...)
100
101 dist.gen.pairs <- function(x,y,...) dist.gen(t(cbind(x,y)),...)
102
103 clustfun <- function(x,clustnr=20,bootnr=50,metric="pearson",do.gap=TRUE,SE.method="Tibs2001SEmax",SE.factor=.25,B.gap=50,cln=0,rseed=17000)
104 {
105 if ( clustnr < 2) stop("Choose clustnr > 1")
106 di <- dist.gen(t(x),method=metric)
107 if ( do.gap | cln > 0 ){
108 gpr <- NULL
109 if ( do.gap ){
110 set.seed(rseed)
111 gpr <- clusGap(as.matrix(di), FUN = kmeans, K.max = clustnr, B = B.gap)
112 if ( cln == 0 ) cln <- maxSE(gpr$Tab[,3],gpr$Tab[,4],method=SE.method,SE.factor)
113 }
114 if ( cln <= 1 ) {
115 clb <- list(result=list(partition=rep(1,dim(x)[2])),bootmean=1)
116 names(clb$result$partition) <- names(x)
117 return(list(x=x,clb=clb,gpr=gpr,di=di))
118 }
119 clb <- clusterboot(di,B=bootnr,distances=FALSE,bootmethod="boot",clustermethod=KmeansCBI,krange=cln,scaling=FALSE,multipleboot=FALSE,bscompare=TRUE,seed=rseed)
120 return(list(x=x,clb=clb,gpr=gpr,di=di))
121 }
122 }
123
124 KmeansCBI <- function (data, krange, k = NULL, scaling = FALSE, runs = 1,
125 criterion = "ch", method="euclidean",...)
126 {
127 if (!is.null(k))
128 krange <- k
129 if (!identical(scaling, FALSE))
130 sdata <- scale(data, center = TRUE, scale = scaling)
131 else sdata <- data
132 c1 <- Kmeansruns(sdata, krange, runs = runs, criterion = criterion, method = method,
133 ...)
134 partition <- c1$cluster
135 cl <- list()
136 nc <- krange
137 for (i in 1:nc) cl[[i]] <- partition == i
138 out <- list(result = c1, nc = nc, clusterlist = cl, partition = partition,
139 clustermethod = "kmeans")
140 out
141 }
142
143 Kmeansruns <- function (data, krange = 2:10, criterion = "ch", iter.max = 100,
144 runs = 100, scaledata = FALSE, alpha = 0.001, critout = FALSE,
145 plot = FALSE, method="euclidean", ...)
146 {
147 data <- as.matrix(data)
148 if (criterion == "asw")
149 sdata <- dist(data)
150 if (scaledata)
151 data <- scale(data)
152 cluster1 <- 1 %in% krange
153 crit <- numeric(max(krange))
154 km <- list()
155 for (k in krange) {
156 if (k > 1) {
157 minSS <- Inf
158 kmopt <- NULL
159 for (i in 1:runs) {
160 options(show.error.messages = FALSE)
161 repeat {
162 kmm <- try(Kmeans(data, k, iter.max = iter.max, method=method,
163 ...))
164 if (class(kmm) != "try-error")
165 break
166 }
167 options(show.error.messages = TRUE)
168 swss <- sum(kmm$withinss)
169 if (swss < minSS) {
170 kmopt <- kmm
171 minSS <- swss
172 }
173 if (plot) {
174 par(ask = TRUE)
175 pairs(data, col = kmm$cluster, main = swss)
176 }
177 }
178 km[[k]] <- kmopt
179 crit[k] <- switch(criterion, asw = cluster.stats(sdata,
180 km[[k]]$cluster)$avg.silwidth, ch = calinhara(data,
181 km[[k]]$cluster))
182 if (critout)
183 cat(k, " clusters ", crit[k], "\n")
184 }
185 }
186 if (cluster1)
187 cluster1 <- dudahart2(data, km[[2]]$cluster, alpha = alpha)$cluster1
188 k.best <- which.max(crit)
189 if (cluster1)
190 k.best <- 1
191 km[[k.best]]$crit <- crit
192 km[[k.best]]$bestk <- k.best
193 out <- km[[k.best]]
194 out
195 }
196
197 binompval <- function(p,N,n){
198 pval <- pbinom(n,round(N,0),p,lower.tail=TRUE)
199 pval[!is.na(pval) & pval > 0.5] <- 1-pval[!is.na(pval) & pval > 0.5]
200 return(pval)
201 }
202
203 setGeneric("clustexp", function(object,clustnr=20,bootnr=50,metric="pearson",do.gap=TRUE,SE.method="Tibs2001SEmax",SE.factor=.25,B.gap=50,cln=0,rseed=17000) standardGeneric("clustexp"))
204
205 setMethod("clustexp",
206 signature = "SCseq",
207 definition = function(object,clustnr,bootnr,metric,do.gap,SE.method,SE.factor,B.gap,cln,rseed) {
208 if ( ! is.numeric(clustnr) ) stop("clustnr has to be a positive integer") else if ( round(clustnr) != clustnr | clustnr <= 0 ) stop("clustnr has to be a positive integer")
209 if ( ! is.numeric(bootnr) ) stop("bootnr has to be a positive integer") else if ( round(bootnr) != bootnr | bootnr <= 0 ) stop("bootnr has to be a positive integer")
210 if ( ! ( metric %in% c( "spearman","pearson","kendall","euclidean","maximum","manhattan","canberra","binary","minkowski") ) ) stop("metric has to be one of the following: spearman, pearson, kendall, euclidean, maximum, manhattan, canberra, binary, minkowski")
211 if ( ! ( SE.method %in% c( "firstSEmax","Tibs2001SEmax","globalSEmax","firstmax","globalmax") ) ) stop("SE.method has to be one of the following: firstSEmax, Tibs2001SEmax, globalSEmax, firstmax, globalmax")
212 if ( ! is.numeric(SE.factor) ) stop("SE.factor has to be a non-negative integer") else if ( SE.factor < 0 ) stop("SE.factor has to be a non-negative integer")
213 if ( ! ( is.numeric(do.gap) | is.logical(do.gap) ) ) stop( "do.gap has to be logical (TRUE/FALSE)" )
214 if ( ! is.numeric(B.gap) ) stop("B.gap has to be a positive integer") else if ( round(B.gap) != B.gap | B.gap <= 0 ) stop("B.gap has to be a positive integer")
215 if ( ! is.numeric(cln) ) stop("cln has to be a non-negative integer") else if ( round(cln) != cln | cln < 0 ) stop("cln has to be a non-negative integer")
216 if ( ! is.numeric(rseed) ) stop("rseed has to be numeric")
217 if ( !do.gap & cln == 0 ) stop("cln has to be a positive integer or do.gap has to be TRUE")
218 object@clusterpar <- list(clustnr=clustnr,bootnr=bootnr,metric=metric,do.gap=do.gap,SE.method=SE.method,SE.factor=SE.factor,B.gap=B.gap,cln=cln,rseed=rseed)
219 y <- clustfun(object@fdata,clustnr,bootnr,metric,do.gap,SE.method,SE.factor,B.gap,cln,rseed)
220 object@kmeans <- list(kpart=y$clb$result$partition, jaccard=y$clb$bootmean, gap=y$gpr)
221 object@distances <- as.matrix( y$di )
222 set.seed(111111)
223 object@fcol <- sample(rainbow(max(y$clb$result$partition)))
224 return(object)
225 }
226 )
227
228 setGeneric("findoutliers", function(object,outminc=5,outlg=2,probthr=1e-3,thr=2**-(1:40),outdistquant=.75) standardGeneric("findoutliers"))
229
230 setMethod("findoutliers",
231 signature = "SCseq",
232 definition = function(object,outminc,outlg,probthr,thr,outdistquant) {
233 if ( length(object@kmeans$kpart) == 0 ) stop("run clustexp before findoutliers")
234 if ( ! is.numeric(outminc) ) stop("outminc has to be a non-negative integer") else if ( round(outminc) != outminc | outminc < 0 ) stop("outminc has to be a non-negative integer")
235 if ( ! is.numeric(outlg) ) stop("outlg has to be a non-negative integer") else if ( round(outlg) != outlg | outlg < 0 ) stop("outlg has to be a non-negative integer")
236 if ( ! is.numeric(probthr) ) stop("probthr has to be a number between 0 and 1") else if ( probthr < 0 | probthr > 1 ) stop("probthr has to be a number between 0 and 1")
237 if ( ! is.numeric(thr) ) stop("thr hast to be a vector of numbers between 0 and 1") else if ( min(thr) < 0 | max(thr) > 1 ) stop("thr hast to be a vector of numbers between 0 and 1")
238 if ( ! is.numeric(outdistquant) ) stop("outdistquant has to be a number between 0 and 1") else if ( outdistquant < 0 | outdistquant > 1 ) stop("outdistquant has to be a number between 0 and 1")
239
240 object@outlierpar <- list( outminc=outminc,outlg=outlg,probthr=probthr,thr=thr,outdistquant=outdistquant )
241 ### calibrate background model
242 m <- log2(apply(object@fdata,1,mean))
243 v <- log2(apply(object@fdata,1,var))
244 f <- m > -Inf & v > -Inf
245 m <- m[f]
246 v <- v[f]
247 mm <- -8
248 repeat{
249 fit <- lm(v ~ m + I(m^2))
250 if( coef(fit)[3] >= 0 | mm >= 3){
251 break
252 }
253 mm <- mm + .5
254 f <- m > mm
255 m <- m[f]
256 v <- v[f]
257 }
258 object@background <- list()
259 object@background$vfit <- fit
260 object@background$lvar <- function(x,object) 2**(coef(object@background$vfit)[1] + log2(x)*coef(object@background$vfit)[2] + coef(object@background$vfit)[3] * log2(x)**2)
261 object@background$lsize <- function(x,object) x**2/(max(x + 1e-6,object@background$lvar(x,object)) - x)
262
263 ### identify outliers
264 out <- c()
265 stest <- rep(0,length(thr))
266 cprobs <- c()
267 for ( n in 1:max(object@kmeans$kpart) ){
268 if ( sum(object@kmeans$kpart == n) == 1 ){ cprobs <- append(cprobs,.5); names(cprobs)[length(cprobs)] <- names(object@kmeans$kpart)[object@kmeans$kpart == n]; next }
269 x <- object@fdata[,object@kmeans$kpart == n]
270 x <- x[apply(x,1,max) > outminc,]
271 z <- t( apply(x,1,function(x){ apply( cbind( pnbinom(round(x,0),mu=mean(x),size=object@background$lsize(mean(x),object)) , 1 - pnbinom(round(x,0),mu=mean(x),size=object@background$lsize(mean(x),object)) ),1, min) } ) )
272 cp <- apply(z,2,function(x){ y <- p.adjust(x,method="BH"); y <- y[order(y,decreasing=FALSE)]; return(y[outlg]);})
273 f <- cp < probthr
274 cprobs <- append(cprobs,cp)
275 if ( sum(f) > 0 ) out <- append(out,names(x)[f])
276 for ( j in 1:length(thr) ) stest[j] <- stest[j] + sum( cp < thr[j] )
277 }
278 object@out <-list(out=out,stest=stest,thr=thr,cprobs=cprobs)
279
280 ### cluster outliers
281 clp2p.cl <- c()
282 cols <- names(object@fdata)
283 di <- as.data.frame(object@distances)
284 for ( i in 1:max(object@kmeans$kpart) ) {
285 tcol <- cols[object@kmeans$kpart == i]
286 if ( sum(!(tcol %in% out)) > 1 ) clp2p.cl <- append(clp2p.cl,as.vector(t(di[tcol[!(tcol %in% out)],tcol[!(tcol %in% out)]])))
287 }
288 clp2p.cl <- clp2p.cl[clp2p.cl>0]
289
290 cpart <- object@kmeans$kpart
291 cadd <- list()
292 if ( length(out) > 0 ){
293 if (length(out) == 1){
294 cadd <- list(out)
295 }else{
296 n <- out
297 m <- as.data.frame(di[out,out])
298
299 for ( i in 1:length(out) ){
300 if ( length(n) > 1 ){
301 o <- order(apply(cbind(m,1:dim(m)[1]),1,function(x) min(x[1:(length(x)-1)][-x[length(x)]])),decreasing=FALSE)
302 m <- m[o,o]
303 n <- n[o]
304 f <- m[,1] < quantile(clp2p.cl,outdistquant) | m[,1] == min(clp2p.cl)
305 ind <- 1
306 if ( sum(f) > 1 ) for ( j in 2:sum(f) ) if ( apply(m[f,f][j,c(ind,j)] > quantile(clp2p.cl,outdistquant) ,1,sum) == 0 ) ind <- append(ind,j)
307 cadd[[i]] <- n[f][ind]
308 g <- ! n %in% n[f][ind]
309 n <- n[g]
310 m <- m[g,g]
311 if ( sum(g) == 0 ) break
312
313 }else if (length(n) == 1){
314 cadd[[i]] <- n
315 break
316 }
317 }
318 }
319
320 for ( i in 1:length(cadd) ){
321 cpart[cols %in% cadd[[i]]] <- max(cpart) + 1
322 }
323 }
324
325 ### determine final clusters
326 for ( i in 1:max(cpart) ){
327 d <- object@fdata[,cols[cpart == i]]
328 if ( sum(cpart == i) == 1 ) cent <- d else cent <- apply(d,1,mean)
329 if ( i == 1 ) dcent <- data.frame(cent) else dcent <- cbind(dcent,cent)
330 if ( i == 1 ) tmp <- data.frame(apply(object@fdata[,cols],2,dist.gen.pairs,y=cent,method=object@clusterpar$metric)) else tmp <- cbind(tmp,apply(object@fdata[,cols],2,dist.gen.pairs,y=cent,method=object@clusterpar$metric))
331 }
332 cpart <- apply(tmp,1,function(x) order(x,decreasing=FALSE)[1])
333
334 for ( i in max(cpart):1){if (sum(cpart==i)==0) cpart[cpart>i] <- cpart[cpart>i] - 1 }
335
336 object@cpart <- cpart
337
338 set.seed(111111)
339 object@fcol <- sample(rainbow(max(cpart)))
340 return(object)
341 }
342 )
343
344 setGeneric("plotgap", function(object) standardGeneric("plotgap"))
345
346 setMethod("plotgap",
347 signature = "SCseq",
348 definition = function(object){
349 if ( length(object@kmeans$kpart) == 0 ) stop("run clustexp before plotgap")
350 plot(object@kmeans$gap)
351 }
352 )
353
354 setGeneric("plotsilhouette", function(object) standardGeneric("plotsilhouette"))
355
356 setMethod("plotsilhouette",
357 signature = "SCseq",
358 definition = function(object){
359 if ( length(object@kmeans$kpart) == 0 ) stop("run clustexp before plotsilhouette")
360 if ( length(unique(object@kmeans$kpart)) < 2 ) stop("only a single cluster: no silhouette plot")
361 kpart <- object@kmeans$kpart
362 distances <- dist.gen(object@distances)
363 si <- silhouette(kpart,distances)
364 plot(si)
365 }
366 )
367
368 setGeneric("plotjaccard", function(object) standardGeneric("plotjaccard"))
369
370 setMethod("plotjaccard",
371 signature = "SCseq",
372 definition = function(object){
373 if ( length(object@kmeans$kpart) == 0 ) stop("run clustexp before plotjaccard")
374 if ( length(unique(object@kmeans$kpart)) < 2 ) stop("only a single cluster: no Jaccard's similarity plot")
375 barplot(object@kmeans$jaccard,names.arg=1:length(object@kmeans$jaccard),ylab="Jaccard's similarity")
376 }
377 )
378
379 setGeneric("plotoutlierprobs", function(object) standardGeneric("plotoutlierprobs"))
380
381 setMethod("plotoutlierprobs",
382 signature = "SCseq",
383 definition = function(object){
384 if ( length(object@cpart) == 0 ) stop("run findoutliers before plotoutlierprobs")
385 p <- object@kmeans$kpart[ order(object@kmeans$kpart,decreasing=FALSE)]
386 x <- object@out$cprobs[names(p)]
387 fcol <- object@fcol
388 for ( i in 1:max(p) ){
389 y <- -log10(x + 2.2e-16)
390 y[p != i] <- 0
391 if ( i == 1 ) b <- barplot(y,ylim=c(0,max(-log10(x + 2.2e-16))*1.1),col=fcol[i],border=fcol[i],names.arg=FALSE,ylab="-log10prob") else barplot(y,add=TRUE,col=fcol[i],border=fcol[i],names.arg=FALSE,axes=FALSE)
392 }
393 abline(-log10(object@outlierpar$probthr),0,col="black",lty=2)
394 d <- b[2,1] - b[1,1]
395 y <- 0
396 for ( i in 1:max(p) ) y <- append(y,b[sum(p <=i),1] + d/2)
397 axis(1,at=(y[1:(length(y)-1)] + y[-1])/2,lab=1:max(p))
398 box()
399 }
400 )
401
402 setGeneric("plotbackground", function(object) standardGeneric("plotbackground"))
403
404 setMethod("plotbackground",
405 signature = "SCseq",
406 definition = function(object){
407 if ( length(object@cpart) == 0 ) stop("run findoutliers before plotbackground")
408 m <- apply(object@fdata,1,mean)
409 v <- apply(object@fdata,1,var)
410 fit <- locfit(v~lp(m,nn=.7),family="gamma",maxk=500)
411 plot(log2(m),log2(v),pch=20,xlab="log2mean",ylab="log2var",col="grey")
412 lines(log2(m[order(m)]),log2(object@background$lvar(m[order(m)],object)),col="red",lwd=2)
413 lines(log2(m[order(m)]),log2(fitted(fit)[order(m)]),col="orange",lwd=2,lty=2)
414 legend("topleft",legend=substitute(paste("y = ",a,"*x^2 + ",b,"*x + ",d,sep=""),list(a=round(coef(object@background$vfit)[3],2),b=round(coef(object@background$vfit)[2],2),d=round(coef(object@background$vfit)[1],2))),bty="n")
415 }
416 )
417
418 setGeneric("plotsensitivity", function(object) standardGeneric("plotsensitivity"))
419
420 setMethod("plotsensitivity",
421 signature = "SCseq",
422 definition = function(object){
423 if ( length(object@cpart) == 0 ) stop("run findoutliers before plotsensitivity")
424 plot(log10(object@out$thr), object@out$stest, type="l",xlab="log10 Probability cutoff", ylab="Number of outliers")
425 abline(v=log10(object@outlierpar$probthr),col="red",lty=2)
426 }
427 )
428
429 setGeneric("clustdiffgenes", function(object,pvalue=.01) standardGeneric("clustdiffgenes"))
430
431 setMethod("clustdiffgenes",
432 signature = "SCseq",
433 definition = function(object,pvalue){
434 if ( length(object@cpart) == 0 ) stop("run findoutliers before clustdiffgenes")
435 if ( ! is.numeric(pvalue) ) stop("pvalue has to be a number between 0 and 1") else if ( pvalue < 0 | pvalue > 1 ) stop("pvalue has to be a number between 0 and 1")
436 cdiff <- list()
437 x <- object@ndata
438 y <- object@expdata[,names(object@ndata)]
439 part <- object@cpart
440 for ( i in 1:max(part) ){
441 if ( sum(part == i) == 0 ) next
442 m <- apply(x,1,mean)
443 n <- if ( sum(part == i) > 1 ) apply(x[,part == i],1,mean) else x[,part == i]
444 no <- if ( sum(part == i) > 1 ) median(apply(y[,part == i],2,sum))/median(apply(x[,part == i],2,sum)) else sum(y[,part == i])/sum(x[,part == i])
445 m <- m*no
446 n <- n*no
447 pv <- binompval(m/sum(m),sum(n),n)
448 d <- data.frame(mean.all=m,mean.cl=n,fc=n/m,pv=pv)[order(pv,decreasing=FALSE),]
449 cdiff[[paste("cl",i,sep=".")]] <- d[d$pv < pvalue,]
450 }
451 return(cdiff)
452 }
453 )
454
455 setGeneric("diffgenes", function(object,cl1,cl2,mincount=5) standardGeneric("diffgenes"))
456
457 setMethod("diffgenes",
458 signature = "SCseq",
459 definition = function(object,cl1,cl2,mincount){
460 part <- object@cpart
461 cl1 <- c(cl1)
462 cl2 <- c(cl2)
463 if ( length(part) == 0 ) stop("run findoutliers before diffgenes")
464 if ( ! is.numeric(mincount) ) stop("mincount has to be a non-negative number") else if ( mincount < 0 ) stop("mincount has to be a non-negative number")
465 if ( length(intersect(cl1, part)) < length(unique(cl1)) ) stop( paste("cl1 has to be a subset of ",paste(sort(unique(part)),collapse=","),"\n",sep="") )
466 if ( length(intersect(cl2, part)) < length(unique(cl2)) ) stop( paste("cl2 has to be a subset of ",paste(sort(unique(part)),collapse=","),"\n",sep="") )
467 f <- apply(object@ndata[,part %in% c(cl1,cl2)],1,max) > mincount
468 x <- object@ndata[f,part %in% cl1]
469 y <- object@ndata[f,part %in% cl2]
470 if ( sum(part %in% cl1) == 1 ) m1 <- x else m1 <- apply(x,1,mean)
471 if ( sum(part %in% cl2) == 1 ) m2 <- y else m2 <- apply(y,1,mean)
472 if ( sum(part %in% cl1) == 1 ) s1 <- sqrt(x) else s1 <- sqrt(apply(x,1,var))
473 if ( sum(part %in% cl2) == 1 ) s2 <- sqrt(y) else s2 <- sqrt(apply(y,1,var))
474
475 d <- ( m1 - m2 )/ apply( cbind( s1, s2 ),1,mean )
476 names(d) <- rownames(object@ndata)[f]
477 if ( sum(part %in% cl1) == 1 ){
478 names(x) <- names(d)
479 x <- x[order(d,decreasing=TRUE)]
480 }else{
481 x <- x[order(d,decreasing=TRUE),]
482 }
483 if ( sum(part %in% cl2) == 1 ){
484 names(y) <- names(d)
485 y <- y[order(d,decreasing=TRUE)]
486 }else{
487 y <- y[order(d,decreasing=TRUE),]
488 }
489 return(list(z=d[order(d,decreasing=TRUE)],cl1=x,cl2=y,cl1n=cl1,cl2n=cl2))
490 }
491 )
492
493 plotdiffgenes <- function(z,gene=g){
494 if ( ! is.list(z) ) stop("first arguments needs to be output of function diffgenes")
495 if ( length(z) < 5 ) stop("first arguments needs to be output of function diffgenes")
496 if ( sum(names(z) == c("z","cl1","cl2","cl1n","cl2n")) < 5 ) stop("first arguments needs to be output of function diffgenes")
497 if ( length(gene) > 1 ) stop("only single value allowed for argument gene")
498 if ( !is.numeric(gene) & !(gene %in% names(z$z)) ) stop("argument gene needs to be within rownames of first argument or a positive integer number")
499 if ( is.numeric(gene) ){ if ( gene < 0 | round(gene) != gene ) stop("argument gene needs to be within rownames of first argument or a positive integer number") }
500 x <- if ( is.null(dim(z$cl1)) ) z$cl1[gene] else t(z$cl1[gene,])
501 y <- if ( is.null(dim(z$cl2)) ) z$cl2[gene] else t(z$cl2[gene,])
502 plot(1:length(c(x,y)),c(x,y),ylim=c(0,max(c(x,y))),xlab="",ylab="Expression",main=gene,cex=0,axes=FALSE)
503 axis(2)
504 box()
505 u <- 1:length(x)
506 rect(u - .5,0,u + .5,x,col="red")
507 v <- c(min(u) - .5,max(u) + .5)
508 axis(1,at=mean(v),lab=paste(z$cl1n,collapse=","))
509 lines(v,rep(mean(x),length(v)))
510 lines(v,rep(mean(x)-sqrt(var(x)),length(v)),lty=2)
511 lines(v,rep(mean(x)+sqrt(var(x)),length(v)),lty=2)
512
513 u <- ( length(x) + 1 ):length(c(x,y))
514 v <- c(min(u) - .5,max(u) + .5)
515 rect(u - .5,0,u + .5,y,col="blue")
516 axis(1,at=mean(v),lab=paste(z$cl2n,collapse=","))
517 lines(v,rep(mean(y),length(v)))
518 lines(v,rep(mean(y)-sqrt(var(y)),length(v)),lty=2)
519 lines(v,rep(mean(y)+sqrt(var(y)),length(v)),lty=2)
520 abline(v=length(x) + .5)
521 }
522
523 setGeneric("clustheatmap", function(object,final=FALSE,hmethod="single") standardGeneric("clustheatmap"))
524
525 setMethod("clustheatmap",
526 signature = "SCseq",
527 definition = function(object,final,hmethod){
528 if ( final & length(object@cpart) == 0 ) stop("run findoutliers before clustheatmap")
529 if ( !final & length(object@kmeans$kpart) == 0 ) stop("run clustexp before clustheatmap")
530 x <- object@fdata
531 part <- if ( final ) object@cpart else object@kmeans$kpart
532 na <- c()
533 j <- 0
534 for ( i in 1:max(part) ){
535 if ( sum(part == i) == 0 ) next
536 j <- j + 1
537 na <- append(na,i)
538 d <- x[,part == i]
539 if ( sum(part == i) == 1 ) cent <- d else cent <- apply(d,1,mean)
540 if ( j == 1 ) tmp <- data.frame(cent) else tmp <- cbind(tmp,cent)
541 }
542 names(tmp) <- paste("cl",na,sep=".")
543 if ( max(part) > 1 ) cclmo <- hclust(dist.gen(as.matrix(dist.gen(t(tmp),method=object@clusterpar$metric))),method=hmethod)$order else cclmo <- 1
544 q <- part
545 for ( i in 1:max(part) ){
546 q[part == na[cclmo[i]]] <- i
547 }
548 part <- q
549 di <- as.data.frame( as.matrix( dist.gen(t(object@distances)) ) )
550 pto <- part[order(part,decreasing=FALSE)]
551 ptn <- c()
552 for ( i in 1:max(pto) ){ pt <- names(pto)[pto == i]; z <- if ( length(pt) == 1 ) pt else pt[hclust(as.dist(t(di[pt,pt])),method=hmethod)$order]; ptn <- append(ptn,z) }
553 col <- object@fcol
554 mi <- min(di,na.rm=TRUE)
555 ma <- max(di,na.rm=TRUE)
556 layout(matrix(data=c(1,3,2,4), nrow=2, ncol=2), widths=c(5,1,5,1), heights=c(5,1,1,1))
557 ColorRamp <- colorRampPalette(brewer.pal(n = 7,name = "RdYlBu"))(100)
558 ColorLevels <- seq(mi, ma, length=length(ColorRamp))
559 if ( mi == ma ){
560 ColorLevels <- seq(0.99*mi, 1.01*ma, length=length(ColorRamp))
561 }
562 par(mar = c(3,5,2.5,2))
563 image(as.matrix(di[ptn,ptn]),col=ColorRamp,axes=FALSE)
564 abline(0,1)
565 box()
566
567 tmp <- c()
568 for ( u in 1:max(part) ){
569 ol <- (0:(length(part) - 1)/(length(part) - 1))[ptn %in% names(x)[part == u]]
570 points(rep(0,length(ol)),ol,col=col[cclmo[u]],pch=15,cex=.75)
571 points(ol,rep(0,length(ol)),col=col[cclmo[u]],pch=15,cex=.75)
572 tmp <- append(tmp,mean(ol))
573 }
574 axis(1,at=tmp,lab=cclmo)
575 axis(2,at=tmp,lab=cclmo)
576 par(mar = c(3,2.5,2.5,2))
577 image(1, ColorLevels,
578 matrix(data=ColorLevels, ncol=length(ColorLevels),nrow=1),
579 col=ColorRamp,
580 xlab="",ylab="",
581 xaxt="n")
582 layout(1)
583 return(cclmo)
584 }
585 )
586
587 setGeneric("comptsne", function(object,rseed=15555) standardGeneric("comptsne"))
588
589 setMethod("comptsne",
590 signature = "SCseq",
591 definition = function(object,rseed){
592 if ( length(object@kmeans$kpart) == 0 ) stop("run clustexp before comptsne")
593 set.seed(rseed)
594 di <- dist.gen(as.matrix(object@distances))
595 ts <- tsne(di,k=2)
596 object@tsne <- as.data.frame(ts)
597 return(object)
598 }
599 )
600
601 setGeneric("plottsne", function(object,final=TRUE) standardGeneric("plottsne"))
602
603 setMethod("plottsne",
604 signature = "SCseq",
605 definition = function(object,final){
606 if ( length(object@tsne) == 0 ) stop("run comptsne before plottsne")
607 if ( final & length(object@cpart) == 0 ) stop("run findoutliers before plottsne")
608 if ( !final & length(object@kmeans$kpart) == 0 ) stop("run clustexp before plottsne")
609 part <- if ( final ) object@cpart else object@kmeans$kpart
610 plot(object@tsne,xlab="Dim 1",ylab="Dim 2",pch=20,cex=1.5,col="lightgrey")
611 for ( i in 1:max(part) ){
612 if ( sum(part == i) > 0 ) text(object@tsne[part == i,1],object@tsne[part == i,2],i,col=object@fcol[i],cex=.75,font=4)
613 }
614 }
615 )
616
617 setGeneric("plotlabelstsne", function(object,labels=NULL) standardGeneric("plotlabelstsne"))
618
619 setMethod("plotlabelstsne",
620 signature = "SCseq",
621 definition = function(object,labels){
622 if ( is.null(labels ) ) labels <- names(object@ndata)
623 if ( length(object@tsne) == 0 ) stop("run comptsne before plotlabelstsne")
624 plot(object@tsne,xlab="Dim 1",ylab="Dim 2",pch=20,cex=1.5,col="lightgrey")
625 text(object@tsne[,1],object@tsne[,2],labels,cex=.5)
626 }
627 )
628
629 setGeneric("plotsymbolstsne", function(object,types=NULL) standardGeneric("plotsymbolstsne"))
630
631 setMethod("plotsymbolstsne",
632 signature = "SCseq",
633 definition = function(object,types){
634 if ( is.null(types) ) types <- names(object@fdata)
635 if ( length(object@tsne) == 0 ) stop("run comptsne before plotsymbolstsne")
636 if ( length(types) != ncol(object@fdata) ) stop("types argument has wrong length. Length has to equal to the column number of object@ndata")
637 coloc <- rainbow(length(unique(types)))
638 syms <- c()
639 plot(object@tsne,xlab="Dim 1",ylab="Dim 2",pch=20,col="grey")
640 for ( i in 1:length(unique(types)) ){
641 f <- types == sort(unique(types))[i]
642 syms <- append( syms, ( (i-1) %% 25 ) + 1 )
643 points(object@tsne[f,1],object@tsne[f,2],col=coloc[i],pch=( (i-1) %% 25 ) + 1,cex=1)
644 }
645 legend("topleft", legend=sort(unique(types)), col=coloc, pch=syms)
646 }
647 )
648
649 setGeneric("plotexptsne", function(object,g,n="",logsc=FALSE) standardGeneric("plotexptsne"))
650
651 setMethod("plotexptsne",
652 signature = "SCseq",
653 definition = function(object,g,n="",logsc=FALSE){
654 if ( length(object@tsne) == 0 ) stop("run comptsne before plottsne")
655 if ( length(intersect(g,rownames(object@ndata))) < length(unique(g)) ) stop("second argument does not correspond to set of rownames slot ndata of SCseq object")
656 if ( !is.numeric(logsc) & !is.logical(logsc) ) stop("argument logsc has to be logical (TRUE/FALSE)")
657 if ( n == "" ) n <- g[1]
658 l <- apply(object@ndata[g,] - .1,2,sum) + .1
659 if (logsc) {
660 f <- l == 0
661 l <- log(l)
662 l[f] <- NA
663 }
664 mi <- min(l,na.rm=TRUE)
665 ma <- max(l,na.rm=TRUE)
666 ColorRamp <- colorRampPalette(rev(brewer.pal(n = 7,name = "RdYlBu")))(100)
667 ColorLevels <- seq(mi, ma, length=length(ColorRamp))
668 v <- round((l - mi)/(ma - mi)*99 + 1,0)
669 layout(matrix(data=c(1,3,2,4), nrow=2, ncol=2), widths=c(5,1,5,1), heights=c(5,1,1,1))
670 par(mar = c(3,5,2.5,2))
671 plot(object@tsne,xlab="Dim 1",ylab="Dim 2",main=n,pch=20,cex=0,col="grey")
672 for ( k in 1:length(v) ){
673 points(object@tsne[k,1],object@tsne[k,2],col=ColorRamp[v[k]],pch=20,cex=1.5)
674 }
675 par(mar = c(3,2.5,2.5,2))
676 image(1, ColorLevels,
677 matrix(data=ColorLevels, ncol=length(ColorLevels),nrow=1),
678 col=ColorRamp,
679 xlab="",ylab="",
680 xaxt="n")
681 layout(1)
682 }
683 )