Mercurial > repos > melpetera > batchcorrection

comparison batchcorrection/Normalisation_QCpool.r @ 2:016780b192a6 draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Uploaded
author melpetera
date Thu, 02 Mar 2017 03:38:08 -0500
parents
children
comparison
equal deleted inserted replaced
1:f64656ae9ea4 2:016780b192a6
1 # Author: jfmartin
2 # Modified by : mpetera
3 ###############################################################################
4 # Correction of analytical effects inter and intra batch on intensities using quality control pooled samples (QC-pools)
5 # according to the algorithm mentioned by Van der Kloet (J Prot Res 2009).
6 # Parameters : a dataframe of Ions intensities and an other of samples? metadata which must contains at least the three following columns :
7 # "batch" to identify the batches of analyses ; need at least 3 QC-pools for linear adjustment and 8 for lo(w)ess adjustment
8 # "injectionOrder" integer defining the injection order of all samples : QC-pools and analysed samples
9 # "sampleType" indicates if defining a sample with "sample" or a QC-pool with "pool"
10 # NO MISSING DATA are allowed
11 # Version 0.91 insertion of ok_norm function to assess correction feasibility
12 # Version 0.92 insertion of slope test in ok_norm
13 # Version 0.93 name of log file define as a parameter of the correction function
14 # Version 0.94 Within a batch, test if all QCpools or samples values = 0. Definition of an error code in ok_norm function (see function for details)
15 # Version 0.99 include non linear lowess correction.
16 # Version 1.00 the corrected result matrix is return transposed in Galaxy
17 # Version 1.01 standard deviation=0 instead of sum of value=0 is used to assess constant data in ok_norm function. Negative values in corrected matrix are converted to 0.
18 # Version 1.02 plotsituation create a result file with the error code of non execution of correction set by function ok_norm
19 # Version 1.03 fix bug in plot with "reg" option. suppression of ok_norm=4 condition if ok_norm function
20 # Version 2.00 Addition of loess function, correction indicator, plots ; modification of returned objects' format, some plots' displays and ok_norm ifelse format
21 # Version 2.01 Correction for pools negative values earlier in norm_QCpool
22 # Version 2.10 Script refreshing ; vocabulary adjustment ; span in parameters for lo(w)ess regression ; conditionning for third line ACP display ; order in loess display
23 # Version 2.11 ok1 and ok2 permutation (ok_norm) ; conditional display of regression (plotsituation) ; grouping of linked lignes + conditioning (normX) ; conditioning for CVplot
24 # Version 2.20 acplight function added from previous toolBox.R [# Version 1.01 "NA"-coding possibility added in acplight function]
25 # Version 2.30 addition of suppressWarnings() for known and controlled warnings ; suppression of one useless "cat" message ; change in Rdata names ; 'batch(es)' in cat
26 # Version 2.90 change in handling of generated negative and Inf values
27 # Version 2.91 Plot improvement
28
29 ok_norm=function(qcp,qci,spl,spi,method) {
30 # Function used for one ion within one batch to determine whether or not batch correction is possible
31 # ok_norm values :
32 # 0 : no preliminary-condition problem
33 # 1 : standard deviation of QC-pools or samples = 0
34 # 2 : insufficient number of QC-pools within a batch (n=3 for linear, n=8 for lowess or loess)
35 # 3 : significant difference between QC-pools' and samples' means
36 # 4 : denominator =0 when on 1 pool per batch <> 0
37 # 5 : (linear regression only) the slopes ratio ?QC-pools/samples? is lower than -0.2
38
39 ok=0
40 if (method=="linear") {minQC=3} else {minQC=8}
41 if (length(qcp)<minQC) { ok=2
42 } else {
43 if (sd(qcp)==0 | sd(spl)==0) { ok=1
44 } else {
45 cvp= sd(qcp)/mean(qcp); cvs=sd(spl)/mean(spl)
46 rttest=t.test(qcp,y=spl)
47 reslsfit=lsfit(qci, qcp)
48 reslsfitSample=lsfit(spl, spi)
49 ordori=reslsfit$coefficients[1]
50 penteB=reslsfit$coefficients[2]
51 penteS=reslsfitSample$coefficients[2]
52 # Significant difference between samples and pools
53 if (rttest$p.value < 0.01) { ok=3
54 } else {
55 # to avoid denominator =0 when on 1 pool per batch <> 0
56 if (method=="linear" & length(which(((penteB*qci)+ordori)==0))>0 ){ ok=6
57 } else {
58 # different sloop between samples and pools
59 if (method=="linear" & penteB/penteS < -0.20) { ok=5 }
60 }}}}
61 ok_norm=ok
62 }
63
64 plotsituation <- function (x, nbid,outfic="plot_regression.pdf", outres="PreNormSummary.txt",fact="batch",span="none") {
65 # Check for all ions in every batch if linear or lo(w)ess correction is possible.
66 # Use ok_norm function and create a file (PreNormSummary.txt) with the error code.
67 # Also create a pdf file with plots of linear and lo(w)ess regression lines.
68 # x: dataframe with ions in columns and samples in rows ; x is the result of concatenation of sample metadata file and ions file
69 # nbid: number of samples description columns (id and factors) with at least : "batch","injectionOrder","sampleType"
70 # outfic: name of regression plots pdf file
71 # fact: factor to be used as categorical variable for plots and PCA.
72 indfact =which(dimnames(x)[[2]]==fact)
73 indtypsamp =which(dimnames(x)[[2]]=="sampleType")
74 indbatch =which(dimnames(x)[[2]]=="batch")
75 indinject =which(dimnames(x)[[2]]=="injectionOrder")
76 lastIon=dim(x)[2]
77 nbi=lastIon-nbid # Number of ions = total number of columns - number of identifying columns
78 nbb=length(levels(x$batch)) # Number of batch = number of levels of "batch" comlumn (factor)
79 nbs=length(x$sampleType[x$sampleType=="sample"])# Number of samples = number of rows with "sample" value in sampleType
80 pdf(outfic,width=27,height=7*ceiling((nbb+2)/3))
81 cat(nbi," ions ",nbb," batch(es) \n")
82 cv=data.frame(matrix(0,nrow=nbi,ncol=2))# initialisation de la dataset qui contiendra les CV
83 pre_bilan=matrix(0,nrow=nbi,ncol=3*nbb) # dataset of ok_norm function results
84 for (p in 1:nbi) {# for each ion
85 par (mfrow=c(ceiling((nbb+2)/3),3),ask=F,cex=1.2)
86 labion=dimnames(x)[[2]][p+nbid]
87 indpool=which(x$sampleType=="pool") # QCpools subscripts in x
88 pools1=x[indpool,p+nbid]; cv[p,1]=sd(pools1)/mean(pools1)# CV before correction
89 for (b in 1:nbb) {# for each batch...
90 xb=data.frame(x[(x$batch==levels(x$batch)[b]),c(indtypsamp,indinject,p+nbid)])
91 indpb = which(xb$sampleType=="pool")# QCpools subscripts in the current batch
92 indsp = which(xb$sampleType=="sample")# samples subscripts in the current batch
93 indbt = which(xb$sampleType=="sample" | xb$sampleType=="pool")# indices de tous les samples d'un batch pools+samples
94 normLinearTest=ok_norm(xb[indpb,3],xb[indpb,2], xb[indsp,3],xb[indsp,2],method="linear")
95 normLoessTest=ok_norm(xb[indpb,3],xb[indpb,2], xb[indsp,3],xb[indsp,2],method="loess")
96 normLowessTest=ok_norm(xb[indpb,3],xb[indpb,2], xb[indsp,3],xb[indsp,2],method="lowess")
97 #cat(dimnames(x)[[2]][p+nbid]," batch ",b," loess ",normLoessTest," linear ",normLinearTest,"\n")
98 pre_bilan[ p,3*b-2]=normLinearTest
99 pre_bilan[ p,3*b-1]=normLoessTest
100 pre_bilan[ p,3*b]=normLowessTest
101 if(length(indpb)>1){
102 if(span=="none"){span1<-1 ; span2<-2*length(indpool)/nbs}else{span1<-span ; span2<-span}
103 resloess=loess(xb[indpb,3]~xb[indpb,2],span=span1,degree=2,family="gaussian",iterations=4,surface="direct")
104 resloessSample=loess(xb[indsp,3]~xb[indsp,2],span=2*length(indpool)/nbs,degree=2,family="gaussian",iterations=4,surface="direct")
105 reslowess=lowess(xb[indpb,2],xb[indpb,3],f=span2)
106 reslowessSample=lowess(xb[indsp,2],xb[indsp,3])
107 liminf=min(xb[indbt,3]);limsup=max(xb[indbt,3])
108 plot(xb[indsp,2],xb[indsp,3],pch=16, main=paste(labion,"batch ",b),ylab="intensity",xlab="injection order",ylim=c(liminf,limsup))
109 points(xb[indpb,2], xb[indpb,3],pch=5)
110 points(cbind(resloess$x,resloess$fitted)[order(resloess$x),],type="l",col="green3")
111 points(cbind(resloessSample$x,resloessSample$fitted)[order(resloessSample$x),],type="l",col="green3",lty=2)
112 points(reslowess,type="l",col="red"); points(reslowessSample,type="l",col="red",lty=2)
113 abline(lsfit(xb[indpb,2],xb[indpb,3]),col="blue")
114 abline(lsfit(xb[indsp,2],xb[indsp,3]),lty=2,col="blue")
115 legend("topleft",c("pools","samples"),lty=c(1,2),bty="n")
116 legend("topright",c("linear","lowess","loess"),lty=1,col=c("blue","red","green3"),bty="n")
117 }
118 }
119 # series de plot avant et apres correction
120 minval=min(x[p+nbid]);maxval=max(x[p+nbid])
121 plot( x$injectionOrder, x[,p+nbid],col=x$batch,ylim=c(minval,maxval),ylab=labion,
122 main=paste0("before correction (CV for pools = ",round(cv[p,1],2),")"))
123 suppressWarnings(plot.design( x[c(indtypsamp,indbatch,indfact,p+nbid)],main="factors effect before correction"))
124 }
125 dev.off()
126 pre_bilan=data.frame(pre_bilan)
127 labion=dimnames(x)[[2]][nbid+1:nbi]
128 for (i in 1:nbb) {
129 dimnames(pre_bilan)[[2]][3*i-2]=paste("batch",i,"linear")
130 dimnames(pre_bilan)[[2]][3*i-1]=paste("batch",i,"loess")
131 dimnames(pre_bilan)[[2]][3*i]=paste("batch",i,"lowess")
132 }
133 bilan=data.frame(labion,pre_bilan)
134 write.table(bilan,file=outres,sep="\t",row.names=F,quote=F)
135 }
136
137
138 normlowess=function (xb,detail="no",vref=1,b,span=NULL) {
139 # Correction function applied to 1 ion in 1 batch. Use a lowess regression computed on QC-pools in order to correct samples intensity values
140 # xb : dataframe for 1 ion in columns and samples in rows.
141 # vref : reference value (average of ion)
142 # b : batch subscript
143 # nbid: number of samples description columns (id and factors) with at least : "batch","injectionOrder","sampleType"
144 indpb = which(xb$sampleType=="pool") # pools subscripts of current batch
145 indsp = which(xb$sampleType=="sample") # samples of current batch subscripts
146 indbt = which(xb$sampleType=="sample" | xb$sampleType=="pool");# batch subscripts of all samples and QC-pools
147 labion=dimnames(xb)[[2]][3]
148 newval=xb[[3]] # initialisation of corrected values = intial values
149 ind <- 0 # initialisation of correction indicator
150 normTodo=ok_norm(xb[indpb,3],xb[indpb,2], xb[indsp,3],xb[indsp,2],method="lowess")
151 #cat("batch:",b," dim xb=",dim(xb)," ok=",normTodo,"\n")
152 if (normTodo==0) {
153 if(length(span)==0){span2<-2*length(indpb)/length(indsp)}else{span2<-span}
154 reslowess=lowess(xb[indpb,2],xb[indpb,3],f=span2) # lowess regression with QC-pools
155 px=xb[indsp,2]; # vector of injectionOrder values only for samples
156 for(j in 1:length(indbt)) {
157 if (xb$sampleType[j]=="pool") {
158 if (reslowess$y[which(indpb==j)]==0) reslowess$y[which(indpb==j)] <- 1
159 newval[j]=(vref*xb[j,3]) / (reslowess$y[which(indpb==j)])}
160 else { # for samples, the correction value cor correspond to the nearest QCpools
161 cor= reslowess$y[which(abs(reslowess$x-px[which(indsp==j)])==min(abs(reslowess$x - px[which(indsp==j)])))]
162 if (length(cor)>1) {cor=cor[1]}
163 if (cor <= 0) {cor=vref} # no modification of initial value
164 newval[j]=(vref*xb[j,3]) / cor
165 }
166 }
167 if (detail=="reg") {
168 liminf=min(xb[indbt,3]);limsup=max(xb[indbt,3])
169 plot(xb[indsp,2],xb[indsp,3],pch=16,main=paste(labion,"batch ",b),ylab="intensity",xlab="injection order",ylim=c(liminf,limsup))
170 points(xb[indpb,2], xb[indpb,3],pch=5)
171 points(reslowess,type="l",col="red")
172 }
173 ind <- 1
174 } else {# if ok_norm <> 0 , we perform a correction based on batch samples average
175 moySample=mean(xb[indsp,3]);if (moySample==0) moySample=1
176 newval[indsp] = (vref*xb[indsp,3])/moySample
177 if(length(indpb)>0){
178 moypool=mean(xb[indpb,3]) ; if (moypool==0) moypool=1
179 newval[indpb] = (vref*xb[indpb,3])/moypool
180 }
181 }
182 newval <- list(norm.ion=newval,norm.ind=ind)
183 return(newval)
184 }
185
186 normlinear <- function (xb,detail="no",vref=1,b,valneg=0) {
187 # Correction function applied to 1 ion in 1 batch.
188 # Use a linear regression computed on QC-pools in order to correct samples intensity values
189 # xb: dataframe with ions in columns and samples in rows; x is a result of concatenation of sample metadata file and ion file
190 # nbid: number of sample description columns (id and factors) with at least "batch", "injectionOrder" and "sampleType"
191 # b: which batch it is
192 # valneg: to determine what to do with generated negative and Inf values
193 indpb = which(xb$sampleType=="pool")# pools subscripts of current batch
194 indsp = which(xb$sampleType=="sample")# samples of current batch subscripts
195 indbt = which(xb$sampleType=="sample" | xb$sampleType=="pool") # QCpools and samples of current batch subscripts
196 labion=dimnames(xb)[[2]][3]
197 newval=xb[[3]] # initialisation of corrected values = intial values
198 ind <- 0 # initialisation of correction indicator
199 normTodo=ok_norm(xb[indpb,3],xb[indpb,2], xb[indsp,3],xb[indsp,2],method="linear")
200 if (normTodo==0) {
201 ind <- 1
202 reslsfit=lsfit(xb[indpb,2],xb[indpb,3]) # linear regression for QCpools
203 reslsfitSample=lsfit(xb[indsp,2],xb[indsp,3]) # linear regression for samples
204 ordori=reslsfit$coefficients[1]
205 pente=reslsfit$coefficients[2]
206 if (detail=="reg") {
207 liminf=min(xb[indbt,3]);limsup=max(xb[indbt,3])
208 plot(xb[indsp,2],xb[indsp,3],pch=16,
209 main=paste(labion,"batch ",b),ylab="intensity",xlab="injection order",ylim=c(liminf,limsup))
210 points(xb[indpb,2], xb[indpb,3],pch=5)
211 abline(reslsfit)
212 abline(reslsfitSample,lty=2)
213 }
214 # correction with rescaling of ion global intensity (vref)
215 newval = (vref*xb[indbt,3]) / (pente * (xb[indbt,2]) + ordori)
216 newval[which((pente * (xb[indbt,2]) + ordori)<1)] <- -1 # to handle cases where 0<denominator<1
217 # handling if any negative values (or null denominators)
218 if(length(which((newval==Inf)|(newval<0)))!=0){
219 toajust <- which((newval==Inf)|(newval<0))
220 if(valneg=="NA"){
221 newval[toajust] <- NA
222 } else {
223 newval[toajust] <- vref * (xb[indbt,3][toajust]) / mean(xb[indbt,3])
224 ### Other possibility
225 ## if(pente>0){ # slope orientation
226 ## newval[toajust]<-(vref*(xb[indbt,3][toajust]))/(pente*ceiling(-ordori/pente+1.00001)+ordori)
227 ## }else{
228 ## newval[toajust]<-(vref*(xb[indbt,3][toajust]))/(pente*floor(-ordori/pente-1.00001)+ordori)
229 ## }
230 }
231 }
232 } else {# if ok_norm!=0 , we perform a correction based on batch samples average.
233 moySample=mean(xb[indsp,3]); if (moySample==0) moySample=1
234 newval[indsp] = (vref*xb[indsp,3])/moySample
235 if(length(indpb)>0){
236 moypool=mean(xb[indpb,3]) ; if (moypool==0) moypool=1
237 newval[indpb] = (vref*xb[indpb,3])/moypool
238 }
239 }
240 newval <- list(norm.ion=newval,norm.ind=ind)
241 return(newval)
242 }
243
244
245 normloess <- function (xb,detail="no",vref=1,b,span=NULL) {
246 # Correction function applied to 1 ion in 1 batch.
247 # Use a loess regression computed on QC-pools in order to correct samples intensity values.
248 # xb : dataframe for 1 ion in columns and samples in rows.
249 # detail : level of detail in the outlog file.
250 # vref : reference value (average of ion)
251 # b : batch subscript
252 indpb = which(xb$sampleType=="pool") # pools subscripts of current batch
253 indsp = which(xb$sampleType=="sample") # samples of current batch subscripts
254 indbt = which(xb$sampleType=="sample" | xb$sampleType=="pool");# batch subscripts of all samples and QCpools
255 labion=dimnames(xb)[[2]][3]
256 newval=xb[[3]] # initialisation of corrected values = intial values
257 ind <- 0 # initialisation of correction indicator
258 normTodo=ok_norm(xb[indpb,3],xb[indpb,2], xb[indsp,3],xb[indsp,2],method="loess")
259 #cat("batch:",b," dim xb=",dim(xb)," ok=",normTodo,"\n")
260 if (normTodo==0) {
261 if(length(span)==0){span1<-1}else{span1<-span}
262 resloess=loess(xb[indpb,3]~xb[indpb,2],span=span1,degree=2,family="gaussian",iterations=4,surface="direct") # loess regression with QCpools
263 cor=predict(resloess,newdata=xb[,2])
264 cor[cor<=1] <- 1
265 newval=(vref*xb[,3]) / cor
266 if(length(which(newval>3*(quantile(newval)[4])))>0){ # in this case no modification of initial value
267 newval <- xb[,3]} else {ind <- 1} # confirmation of correction
268 if ((detail=="reg")&(ind==1)) { # plot
269 liminf=min(xb[indbt,3]);limsup=max(xb[indbt,3])
270 plot(xb[indsp,2],xb[indsp,3],pch=16,main=paste(labion,"batch ",b),ylab="intensity",xlab="injection order",ylim=c(liminf,limsup))
271 points(xb[indpb,2], xb[indpb,3],pch=5)
272 points(cbind(resloess$x,resloess$fitted)[order(resloess$x),],type="l",col="red")
273 }
274 }
275 if (ind==0) {# if ok_norm != 0 or if correction creates outliers, we perform a correction based on batch samples average
276 moySample=mean(xb[indsp,3]);if (moySample==0) moySample=1
277 newval[indsp] = (vref*xb[indsp,3])/moySample
278 if(length(indpb)>0){
279 moypool=mean(xb[indpb,3]) ; if (moypool==0) moypool=1
280 newval[indpb] = (vref*xb[indpb,3])/moypool
281 }
282 }
283 newval <- list(norm.ion=newval,norm.ind=ind)
284 return(newval)
285 }
286
287
288
289 norm_QCpool <- function (x, nbid, outlog, fact, metaion, detail="no", NormMoyPool=F, NormInt=F, method="linear",span="none",valNull="0")
290 {
291 ### Correction applying linear or lo(w)ess correction function on all ions for every batch of a dataframe.
292 # x: dataframe with ions in column and samples' metadata
293 # nbid: number of sample description columns (id and factors) with at least "batch", "injectionOrder", "sampleType"
294 # outlog: name of regression plots and PCA pdf file
295 # fact: factor to be used as categorical variable for plots
296 # metaion: dataframe of ions' metadata
297 # detail: level of detail in the outlog file. detail="no" ACP + boxplot of CV before and after correction.
298 # detail="plot" with plot for all batch before and after correction.
299 # detail="reg" with added plots with regression lines for all batches.
300 # NormMoyPool: not used
301 # NormInt: not used
302 # method: regression method to be used to correct : "linear" or "lowess" or "loess"
303 # valNull: to determine what to do with negatively estimated intensities
304 indfact =which(dimnames(x)[[2]]==fact)
305 indtypsamp=which(dimnames(x)[[2]]=="sampleType")
306 indbatch =which(dimnames(x)[[2]]=="batch")
307 indinject =which(dimnames(x)[[2]]=="injectionOrder")
308 lastIon=dim(x)[2]
309 valref=apply(as.matrix(x[,(nbid+1):(lastIon)]),2,mean) # reference value for each ion used to still have the same rought size of values
310 nbi=lastIon-nbid # number of ions
311 nbb=length(levels(x$batch)) # Number of batch(es) = number of levels of factor "batch" (can be =1)
312 nbs=length(x$sampleType[x$sampleType=="sample"])# Number of samples
313 nbp=length(x$sampleType[x$sampleType=="pool"])# Number of QCpools
314 Xn=data.frame(x[,c(1:nbid)],matrix(0,nrow=nbp+nbs,ncol=nbi))# initialisation of the corrected dataframe (=initial dataframe)
315 dimnames(Xn)=dimnames(x)
316 cv=data.frame(matrix(0,nrow=nbi,ncol=2))# initialisation of dataframe containing CV before and after correction
317 dimnames(cv)[[2]]=c("avant","apres")
318 if (detail!="reg" && detail!="plot" && detail!="no") {detail="no"}
319 pdf(outlog,width=27,height=20)
320 cat(nbi," ions ",nbb," batch(es) \n")
321 if (detail=="plot") {if(nbb<6){par(mfrow=c(3,3),ask=F,cex=1.5)}else{par(mfrow=c(4,4),ask=F,cex=1.5)}}
322 res.ind <- matrix(NA,ncol=nbb,nrow=nbi,dimnames=list(dimnames(x)[[2]][-c(1:nbid)],paste("norm.b",1:nbb,sep="")))
323 for (p in 1:nbi) {# for each ion
324 labion=dimnames(x)[[2]][p+nbid]
325 if (detail == "reg") {if(nbb<6){par(mfrow=c(3,3),ask=F,cex=1.5)}else{par(mfrow=c(4,4),ask=F,cex=1.5)}}
326 indpool=which(x$sampleType=="pool")# QCpools subscripts in all batches
327 pools1=x[indpool,p+nbid]; cv[p,1]=sd(pools1)/mean(pools1)# CV before correction
328 for (b in 1:nbb) {# for every batch
329 indpb = which(x$batch==levels(x$batch)[b] & x$sampleType=="pool")# QCpools subscripts of the current batch
330 indsp = which(x$batch==levels(x$batch)[b] & x$sampleType=="sample")# samples subscripts of the current batch
331 indbt = which(x$batch==levels(x$batch)[b] & (x$sampleType=="pool" | x$sampleType=="sample")) # subscripts of all samples
332 # cat(dimnames(x)[[2]][p+nbid]," indsp:",length(indsp)," indpb=",length(indpb)," indbt=",length(indbt)," ")
333 sub=data.frame(x[(x$batch==levels(x$batch)[b]),c(indtypsamp,indinject,p+nbid)])
334 if (method=="linear") { res.norm = normlinear(sub,detail,valref[p],b,valNull)
335 } else { if (method=="loess"){ res.norm <- normloess(sub,detail,valref[p],b,span)
336 } else { if (method=="lowess"){ res.norm <- normlowess(sub,detail,valref[p],b,span)
337 } else {stop("\n--\nNo valid 'method' argument supplied.\nMust be 'linear','loess' or 'lowess'.\n--\n")}
338 }}
339 Xn[indbt,p+nbid] = res.norm[[1]]
340 res.ind[p,b] <- res.norm[[2]]
341 # CV batch test : if after normaliszation, CV before < CV after initial values are kept
342 # moypoolRaw=mean(x[indpb,p+nbid]) ; if (moypoolRaw==0) moypoolRaw=1
343 # moySampleRaw=mean(x[indsp,p+nbid]); if (moySampleRaw==0) moySampleRaw=1
344 # moypool=mean(Xn[indpb,p+nbid]) ; if (moypool==0) moypool=1
345 # #moySample=mean(Xn[indsp,p+nbid]); if (moySample==0) moySample=1
346 # if (sd( Xn[indpb,p+nbid])/moypool>sd(x[indpb,p+nbid])/moypoolRaw) {
347 # Xn[indpb,p+nbid] = (valref[p]*x[indpb,p+nbid])/moypoolRaw
348 # Xn[indsp,p+nbid] = (valref[p]*x[indsp,p+nbid])/moySampleRaw
349 # }
350 }
351 Xn[indpool,p+nbid][Xn[indpool,p+nbid]<0] <- 0
352 pools2=Xn[indpool,p+nbid]; cv[p,2]=sd(pools2,na.rm=TRUE)/mean(pools2,na.rm=TRUE)# CV apres correction
353 if (detail=="reg" || detail=="plot" ) {
354 # plot before and after correction
355 minval=min(cbind(x[p+nbid],Xn[p+nbid]),na.rm=TRUE);maxval=max(cbind(x[p+nbid],Xn[p+nbid]),na.rm=TRUE)
356 plot( x$injectionOrder, x[,p+nbid],col=x$batch,ylab=labion,ylim=c(minval,maxval),
357 main=paste0("before correction (CV for pools = ",round(cv[p,1],2),")"))
358 points(x$injectionOrder[indpool],x[indpool,p+nbid],col="maroon",pch=".",cex=2)
359 plot(Xn$injectionOrder,Xn[,p+nbid],col=x$batch,ylab="",ylim=c(minval,maxval),
360 main=paste0("after correction (CV for pools = ",round(cv[p,2],2),")"))
361 points(Xn$injectionOrder[indpool],Xn[indpool,p+nbid],col="maroon",pch=".",cex=2)
362 suppressWarnings(plot.design( x[c(indtypsamp,indbatch,indfact,p+nbid)],main="factors effect before correction"))
363 suppressWarnings(plot.design(Xn[c(indtypsamp,indbatch,indfact,p+nbid)],main="factors effect after correction"))
364 }
365 }
366 ### Replacement of post correction negative values by chosen value
367 Xnn=Xn
368 for (i in c((nbid+1):dim(Xn)[2])) {
369 cneg=which(Xn[[i]]<0)
370 Xnn[[i]]=replace(Xn[[i]],cneg,as.numeric(valNull))
371 }
372 Xn=Xnn
373
374 if (detail=="reg" || detail=="plot" || detail=="no") {
375 if (nbi > 3) {
376 par(mfrow=c(3,4),ask=F,cex=1.2) # PCA Plot before/after, normed only and ions plot
377 acplight(x[,c(indtypsamp,indbatch,indtypsamp,indfact,(nbid+1):lastIon)],"uv",TRUE)
378 norm.ion <- which(colnames(Xn)%in%(rownames(res.ind)[which(rowSums(res.ind)>=1)]))
379 acplight(Xn[,c(indtypsamp,indbatch,indtypsamp,indfact,(nbid+1):lastIon)],"uv",TRUE,norm.ion)
380 if(length(norm.ion)>0){acplight(Xn[,c(indtypsamp,indbatch,indtypsamp,indfact,norm.ion)],"uv",TRUE)}
381 par(mfrow=c(1,2),ask=F,cex=1.2) # Before/after boxplot
382 cvplot=cv[!is.na(cv[[1]])&!is.na(cv[[2]]),]
383 if(nrow(cvplot)>0){
384 boxplot(cvplot[[1]],ylim=c(min(cvplot),max(cvplot)),main="CV before correction")
385 boxplot(cvplot[[2]],ylim=c(min(cvplot),max(cvplot)),main="CV after correction")
386 }
387 dev.off()
388 }
389 }
390 if (nbi<=3) {dev.off()}
391 # transposed matrix is return (format of the initial matrix with ions in rows)
392 Xr=Xn[,-c(1:nbid)]; dimnames(Xr)[[1]]=Xn[[1]]
393 Xr=t(Xr) ; Xr <- data.frame(ions=rownames(Xr),Xr)
394
395 res.norm[[1]] <- Xr ; res.norm[[2]] <- data.frame(metaion,res.ind) ; res.norm[[3]] <- x[,c(1:nbid)]
396 names(res.norm) <- c("dataMatrix","variableMetadata","sampleMetadata")
397 return(res.norm)
398 }
399
400
401
402
403
404 acplight <- function(ids, scaling="uv", indiv=FALSE,indcol=NULL) {
405 suppressPackageStartupMessages(library(ade4))
406 suppressPackageStartupMessages(library(pcaMethods))
407 # Make a PCA and plot scores and loadings.
408 # First column must contain samples' identifiers.
409 # Columns 2 to 4 contain factors to colour the plots.
410 for (i in 1:3) {
411 idss=ids[which(ids[,i+1]!="NA"),]
412 idss=data.frame(idss[idss[,i+1]!="",])
413 classe=as.factor(idss[[i+1]])
414 idsample=as.character(idss[[1]])
415 colour=1:length(levels(classe))
416 ions=as.matrix(idss[,5:dim(idss)[2]])
417 # Removing ions containing NA (not compatible with standard PCA)
418 ions=t(na.omit(t(ions)))
419 if(i==1){if(ncol(ions)!=(ncol(idss)-4)){cat("Note:",(ncol(idss)-4)-ncol(ions),"ions were ignored for PCA display due to NA in intensities.\n")}}
420 # Scaling choice: "uv","none","pareto"
421 object=suppressWarnings(prep(ions, scale=scaling, center=TRUE))
422 if(i==1){if(length(which(apply(ions,2,var)==0))>0){cat("Warning: there are",length(which(apply(ions,2,var)==0)),"constant ions.\n")}}
423 # ALGO: nipals,svdImpute, Bayesian, svd, probalistic=F
424 result <- pca(object, center=F, method="svd", nPcs=2)
425 # ADE4 : to plot samples' ellipsoid for each class
426 s.class(result@scores, classe, cpoint = 1,xax=1,yax=2,col=colour,sub=sprintf("Scores - PCs %sx%s",1,2), possub="bottomright")
427 #s.label(result@loadings,label = ions, cpoint = 0, clabel=0.4, xax=1,yax=2,sub="Loadings",possub="bottomright")
428 if(i==1){resulti <- result}
429 }
430 if(indiv) {
431 colour <- rep("darkblue",length(resulti@loadings)) ; if(!is.null(indcol)) {colour[-c(indcol)] <- "red"}
432 plot(resulti@loadings,col=colour,main="Loadings",xaxt="n",yaxt="n",pch=20,
433 xlab=bquote(PC1-R^2==.(resulti@R2[1])),ylab=bquote(PC2 - R^2 == .(resulti@R2[2])))
434 abline(h=0,v=0)}
435 }
436
437