-
Notifications
You must be signed in to change notification settings - Fork 0
/
CMAFunctions.R
936 lines (838 loc) · 35.1 KB
/
CMAFunctions.R
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
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
#Pin block normalization functions
#Mark Dane
#Version 1
#Version 1.1 adding replicate handling
medianBlock<-function(x,drawPlots=FALSE){
#A low level normalization function
#x is a matrix of intensity values
#x is returned centered about its median
x<-x-median(x, na.rm=TRUE)
if(drawPlots){
plot(x, main="Median Fit")
}
return(x)
}
loessBlock<-function(ints, drawPlots=FALSE, span=.1){
#A low level normalization function
#ints is a matrix of intensity values
#order is a print order that may be used
#the default is to order the values by row
#The residuals of the actual values from the loess fitted values are returned.
z<-as.numeric(t(ints))
cols<-1:dim(ints)[2]
rows<-1:dim(ints)[1]
x<-rep(cols,times=(dim(ints)[1]))
y<-rep(rows,each=(dim(ints)[2]))
loess.model<-loess(z~x*y,na.action='na.exclude', span=span)
residuals<-z-predict(loess.model)
if(drawPlots){
image(y=cols,x=rows,matrix(predict(loess.model),nrow=length(rows),byrow=TRUE))
# plot(x[order(loessOrder)], main="Loess Fit")
# lines(predict(loess.model,order)[order(loessOrder)], col='red')
# abline(median(x, na.rm=TRUE),b=0, col="blue")
}
return(residuals)
}
loessPlate<-function(ints, drawPlots=FALSE, span=.1){
#A low level normalization function
#ints is a matrix of intensity values
#The residuals of the actual values from the loess fitted values are returned.
z<-as.numeric(t(ints))
cols<-1:dim(ints)[2]
rows<-1:dim(ints)[1]
x<-rep(cols,times=(dim(ints)[1]))
y<-rep(rows,each=(dim(ints)[2]))
loess.model<-loess(z~x*y,na.action='na.exclude', span=span)
residuals<-z-predict(loess.model)
# if(drawPlots){
# plot(x[order(loessOrder)], main="Loess Fit")
# lines(predict(loess.model,order)[order(loessOrder)], col='red')
# abline(median(x, na.rm=TRUE),b=0, col="blue")
# }
return(residuals)
}
pinBlock<-function(x,order=1:length(x), method='median', drawPlots=FALSE){
#A pin block normalization wrapper
#x is a matrix to be normalized. x will be read in by row first.
#order is vector that orders the matrix for some normalizations. For instance,
# the print order may be used for the loess normalization.
#y must be a numeric vector with the same length as x
#method is a character specifying the normalization method. Allowed values are "median"
# "loess"
switch(EXPR = method, median=medianBlock(x,drawPlots), loess=loessBlock(x,order,drawPlots))
}
refChannelNormalize<-function(x,refChannel, logged=FALSE){
#Return an object with channels normalized by dividing by the reference channel
#Read the number of channels from the object
nrChannels<-dim(Data(x))[3]
for(ch in 1:nrChannels){
if(ch!=refChannel){
if(!logged) Data(x)[,,ch]<-Data(x)[,,ch]/(Data(x)[,,refChannel]+0.001)
else Data(x)[,,ch]<-Data(x)[,,ch]-(Data(x)[,,refChannel])
}
}
x@state["normalized"]<-TRUE
return(x)
}
pinTipNormalize<-function(x,order=1:pdim(x)[1],method='median',drawPlots=FALSE){
#pin tip normalize the cellHTS object x using method (only median and loess are supported)
#browser()
#Read grid, row and column from the object
grids<-fData(x)[,"Grid"]
rows<-fData(x)[,"Row"]
cols<-fData(x)[,"Column"]
nrReplicates<-dim(Data(x))[2]
#Read the number of wells in each plate from the object
nrWell <- prod(pdim(x))
#Read the number of plates in the screen from the object
nrPlates <- max(plate(x))
#Compute the number of blocks from the objects position data
nrGrids<-max(grids)
#Read the number of channels from the object
nrChannels<-dim(Data(x))[3]
#Extract the configured and annotated assay data into a data frame and add the plate and position data
xAssayData<-Data(x)
xAssay<-data.frame(Data(x))
xAssay<-cbind(xAssay,plate=plate(x),grids=grids,blockRow=rows,blockCols=cols,order=order)
#Create a boxplot of the raw pin block intensities of channel 1
#boxplot(xAssay$X1.intensity~blocks, data=xAssay)
#Normalize each channel and replicate in each block in each plate
for(p in 1:nrPlates) {
for(chr in 1:(nrChannels*nrReplicates)){
for(b in 1:nrGrids){
#browser()
xAssay[,chr][xAssay$plate==p & xAssay$grids==b]<-
pinBlock(xAssay[,chr][xAssay$plate==p & xAssay$grids==b],order, method,drawPlots)
}
}
}
# }
# }
#Put the normalized data back into the original array
#This replaces each channel separately. Should be upgraded with a more compact method
for (ch in 1:nrChannels){
for(rep in 1:nrReplicates){
xAssayData[,rep,ch]<-xAssay[,((ch-1)*nrReplicates+rep)]
}
}
#Put the array with the normalized data back into the new cellHTS object
xpt<-x
Data(xpt)<-xAssayData
xpt@state["normalized"]<-TRUE
return(xpt)
}
#TODO: create a function that takes normalized and scored objects, creates a toptable
# and dedups it writing it to a file based on its name and returning a toptable dataframe
naiveScoreReplicates<-function(xsc){
#Naive Replicate Scoring
#Author: Mark Dane
#Get the scores and gene ID's from the scored object
xscData<-Data(xsc)
xscfData<-fData(xsc)
#Find all instances of the gene in the object and average their scores
for(gene in xscfData$GeneID){
same<-which(xscfData$GeneID %in% gene)
score<-mean(xscData[same])
xscData[same]<-score
}
#Put the averaged scores back into the object
Data(xsc)<-xscData
#Return the scored object with average replicate scores
return(xsc)
}
wellList<-function(xsc,thresh=c(-2.92,2.92)){
#Make a dataframe from the assay and feature data
xscd<-Data(xsc)
xscf<-fData(xsc)
out<-data.frame(xscf,xscd,seqPos=1:(length(xscd)/as.numeric(max(xscf$plate))))
names(out)<-c('plate','well','finalWellAnno','siRNAID', 'GeneID','posName','score','position')
#Eliminate genes with NA scores
out<-out[!is.na(out$score),]
#Identify which wells have duplicate genes and eliminate them
dup<-duplicated(out$GeneID)
outUnique<-out[which(!dup),]
#Create a file for input to the scan^R
#File format is: GeneID score well wellPos Plate finalWellAnno
#Rearrange the output dataframe to create the file format order
tempDF<-with(outUnique,data.frame(GeneID, score, well, wellPos=position,
plate, finalWellAnno, stringsAsFactors=FALSE))
#Sort the output by scores
tempDF<-tempDF[order(tempDF$score, decreasing=TRUE),]
#Apply the thresholds
tempDF<-tempDF[(tempDF$score<=thresh[1]) | (tempDF$score>=thresh[2]),]
#Write the dataframe to a tab delimited file with a header row
write.table(tempDF,file='wellSelect.txt', sep="\t", row.names=FALSE, quote=FALSE)
return(tempDF)
}
readBackwards<-function(x){
#flip the columns in a dataframe up down and left right to make up for
#an error in how the array-pro reads the data
for(i in 1:dim(x)[2]){
#reverse the order of the rows
x[,i]<-x[dim(x)[1]:1,i]
}
return(x)
}
getANGrid<-function(grid, row, col, nrGridCols, nrRows, nrCols){
#given vectors (grid, row, col) of well coordinates, the number of rows and columns in a grid (nrRows,nrcols),
#and the number of grids in a CSMA row (nrGrids) this function returns the corresponding alhpanumeric notation
plateRow<-as.integer((grid-1)/nrGridCols)*nrRows+row
plateCol<-as.integer((grid-1)%%nrGridCols)*nrCols+col
return(unname(unlist(getAlphaNumeric(plateRow,plateCol)[1])))
}
logObject<-function(x,base=2){
#Returns the cellHTS2 object with the Data slot replaced with the log of the values
#TODO: make this a robust log function to deal with non-positive values
Data(x)<-log(Data(x), base=base)
Data(x)[Data(x)<=-1e10]<-NA
return(x)
}
selectColumn<-function(wd=getwd(),infile,outfile,sel=1){
x<-read.table(file=paste(wd,infile,sep=""), header=TRUE, sep="\t",stringsAsFactors=FALSE)
write.table(cbind(x[1],x[,sel],x[,dim(x)[2]]),
file=paste(wd,outfile, sep=""),
sep="\t", quote=FALSE,col.names = c("\t","intensities","position"))
}
orderDF<-function(df,blockGridCols=4){
#Create an order array that converts from block, row, column space to subarray row and column space
#This is used for GAL and aray pro files that are organized by block, row, columns
#Return the dataframe in the correct space
blockGridRows<-max(df$Grid)/blockGridCols
blockRows<-max(df$Row)
blockCols<-max(df$Column)
blockSize<-blockRows*blockCols
nrBlocks<-blockGridRows*blockGridCols
nrWells<-nrBlocks*blockSize
df<-cbind(df,Index=1:nrWells)
#Create ordering pin block matrices.
outDF<-df
for(blockGridRow in 0:(blockGridRows-1)){
blocks<-(((blockGridRow)*blockGridCols)+1):(((blockGridRow)*blockGridCols)+blockGridCols)
selectBs<-df[df$Grid %in% blocks,]
ord<-selectBs$Index[order(selectBs$Row,selectBs$Grid,selectBs$Column)]
outDF[df$Grid %in% blocks,]<-df[ord,]
}
return(outDF)
}
orderDFV2<-function(df,gridsPerRow=4){
#Convert a dataframe from block, row, column space to subarray row and column space
#This is used for GAL and aray pro files that are organized by block, row, columns
#Return the dataframe in the correct space
rowsPerGrid<-max(df$Row)
colsPerGrid<-max(df$Column)
df$arrayGridRow<-ceiling(df$Grid/gridsPerRow)
df$arrayRow<-(df$arrayGridRow-1)*rowsPerGrid+df$Row
df$arrayColumn<-((df$Grid-1)%%gridsPerRow)*colsPerGrid+df$Column
df<-df[order(df$arrayRow,df$arrayColumn),]
df<-subset(df,select=-c(arrayGridRow))
df$PlateIndex<-1:(max(df$arrayColumn)*max(df$arrayRow))
return(df)
}
pseudoPrintOrder<-function(blockRows=10,blockCols=10,blockGridRows=12,blockGridCols=4){
#return an array of a pseudo print order that is sequential within each pin grid
blockSize<-blockRows*blockCols
nrBlocks<-blockGridRows*blockGridCols
nrWells<-nrBlocks*blockRows*blockCols
#Create ordering pin block matrices
blocks<-list()
for(block in 1:nrBlocks){
blocks[[block]]<-matrix(1:blockSize, ncol=blockCols, byrow=TRUE)
}
#Combine 4 pin block matrices to form a row
orderMat<-cbind(blocks[[1]], blocks[[2]],
blocks[[3]], blocks[[4]])
#Combine row groups into a matrix that matches the array
for(blockGridRow in 2:blockGridRows){
index<-(blockGridRow-1)*blockGridCols+1
blockRow<-cbind(blocks[[index]], blocks[[index+1]],
blocks[[index+2]], blocks[[index+3]])
orderMat<-rbind(orderMat,blockRow)
}
#Read the matrix out by row into an ordering array
orderArray<-array(t(orderMat))
return(orderArray)
}
orderArray<-function(positionRows,positionColumns,blockGridRows=12,blockGridCols=4){
#Create an order array that converts from block, row, column space to subarray row and column space
#This is used for GAL file annotations that are organized by block, row, columns
#Using the output of this function to order a GAL file will put order it by rows and then columns
#Caution on R's default of reading by column then row.
#Usage restriction: this function assumes there are four pin grids across a subarray row
#TODO: Make this general for rows that don't have 4 grids columns
#TODO Generalize this for the number of plates
blockRows<-max(positionRows)
blockCols<-max(positionColumns)
blockSize<-blockRows*blockCols
nrBlocks<-blockGridRows*blockGridCols
nrWells<-nrBlocks*blockSize
#Create ordering pin block matrices.
blocks<-list()
for(block in 1:nrBlocks){
blocks[[block]]<-matrix(((block-1)*blockSize+1):((block-1)*blockSize+blockSize),
ncol=blockCols, byrow=TRUE)
}
#Combine 4 pin block matrices to form a row
orderMat<-cbind(blocks[[1]], blocks[[2]],
blocks[[3]], blocks[[4]])
#Combine row groups into a matrix that matches the array
for(blockGridRow in 2:blockGridRows){
index<-(blockGridRow-1)*blockGridCols+1
blockRow<-cbind(blocks[[index]], blocks[[index+1]],
blocks[[index+2]], blocks[[index+3]])
orderMat<-rbind(orderMat,blockRow)
}
#Read the matrix out by row into an ordering array
orderArray<-array(t(orderMat))
orderArray<-c(orderArray, nrWells+orderArray)
return(orderArray)
}
logFileFUN<-function(x,thresh=2){
#Return a dataframe for the log file that has the spots with cell counting channel (3)
#readings below a threshold. These will be flagged in the downstream analysis
#Ouput format to include Plate, Well, Flag, Channel of the spots less than thresh
#Get the data and feature data from the object. Returns as a 3D matrix
#dim 1 is the rows
#dim 2, ? and set to 1 for the debugging example
#dim 3 has a dimension for each channel of data
#TODO: Get this function to stop thowing warnings about the row names
xData<-Data(x)
xfData<-fData(x)
#Create an empty data frame
logDF<-data.frame()
channel<-dim(Data(x))[3]
#Try to use the channel data directly in xData to build the output dataframe
#Cycle through all four replicates which are stored in the second dimension of xData
for(rep in 1:dim(xData)[2]){
#find spots in channel within the flagging range
#Don't attempt to make a dataframe if there are no values to be flagged
if(length(which(xData[,rep,channel]<=thresh)) | sum(is.na(xData[,rep,channel]))){
#Found at least one value in the flagging range so create a dataframe of all spots found
foo<-data.frame(Plate=1,Sample=rep,
Well=xfData$well[which(xData[,rep,channel]<=thresh | is.na(xData[,rep,channel]))],
Flag="NA",Channel=channel,
stringsAsFactors=FALSE,row.names=NULL)
#add the current spots to any already found
logDF<-rbind(logDF,foo)
}#End current replicate
}#End for(rep in 1:dim(x)[2]){
return(logDF)
}#end LogFileFUn
spatialVariation<-function(x){
#Do a spatial variation analysis of the raw data of each channel in a cellHT object that has been
#normalized with the locfit method.
#Get the locfit model of the raw data
#######Hard coded this only works for cellHTS objects with one plate
nrPlates<-length(x@plateData)
nrChannels<-dim(Data(x))[3]
nrReplicates<-dim(x@plateData[[1]])[2]
#Do a spatial variation analysis of the raw data
#Get the locfit model of the raw data
pe<-plateEffects(x)
####Hard coded for 1 plate #####Get the residuals of the loc fit model from its median
#This create a deviation value of the model from the plate median for every spot/well
#This value should capture the magnitude of larger spatial effects while ignoring smaller ones
#The nn parameter in locfit controls the sensitivity between large and small spatial effects
spatialVarsReps<-sapply(X=1:nrReplicates,FUN=function(rep,x,pe,nrChannels){
#Get the total variation of the model from it's median
peChRes<-sapply(X=1:nrChannels,FUN=function(ch,pe,rep){
#Get the variation of all channels in the current replicate
#as a proportion of the median
#Get a logical vector of the non-empty wells
nonEmpty<-matrix(data=x@plateConf$Content!="empty",nrow=dim(x)[1],byrow=FALSE)
#Remove the empty wells from pe and the calculation of Spatial Variation
peRepCh<-pe$rowcol[nonEmpty[,rep],rep,ch]
return(sum(abs(peRepCh-median(peRepCh))/median(peRepCh)))
},pe=pe,rep=rep)
},x=x,pe=pe,nrChannels=nrChannels)
return(spatialVarsReps/dim(x)[1])
}
getRawExcelDF<-function(file, dataType="Net",...){
#Read an excel file into a dataframe
temp<-read.xls(file, verbose=FALSE, stringsAsFactors=FALSE,check.names=TRUE,...)
#Find the dataType columns
dataCols<-grep(pattern=dataType,x=names(temp))
#Find the pin grid columns
annotationCols<-which(names(temp) %in% c("Grid","Row","Column"))
#Work around bug in read.xls that reads some numbers as characters
df<-data.frame(unlist(apply(X=temp[,c(dataCols,annotationCols)],
MARGIN=2,FUN=as.numeric)))
#Remove statistic rows after the data
maxRow<-which(temp$X=="Maximum")
df<-df[1:(df$Grid[maxRow]*df$Row[maxRow]*df$Column[maxRow]),]
rm(temp)
#Make meaningful short names for the columns of the channel number,
#wavelength and dataType
for(col in 1:length(dataCols)){
#find the wavelength in the column name
m <- regexpr(pattern="488|532|635", text=names(df[col]))
#Get the wavelength from the column name
wv<-regmatches(names(df[col]), m)
names(df)[col]<-paste0("Ch",col,"_",wv,"_",dataType)
}
return(df)
}#End getRawexcelDF
# Gamma distribution approximation of p value for rank product.
#Author R.Eisinga et al./FEBS Letters 587 (2013) 677-682
righttailgamma = function(r,k,n) 1 - pgamma(-log(r/(n+1)^k),k,scale=1)
rbindDfs<-function(dfs){
#rbind the replicates into a single dataframe and add the replicate column
out<-dfs[[1]]
for(df in dfs[-1]){
out<-rbind(out,df)
}
out$replicate<-rep(x=1:length(dfs),each=dim(dfs[[1]])[1])
return(out)
}#End rbindDFs
readReplicates<-function(dataFile,gal){
#read the datafile and merge it with the gal annotations
#Read the excel spreadsheet into a dataframe
values<-read.xls(dataFile, verbose=FALSE, stringsAsFactors=FALSE)
#Change the Grid column to numeric
values$Grid<-as.numeric(values$Grid)
#merge the gal and values dataframes on Grid, Row and Column
values<-values[1:max(suppressWarnings(as.double(values$X)), na.rm=TRUE),]
sheetDF<-merge(x=gal,y=values,by=c("Grid","Row","Column"))
#Order the input dataframes by row then column
sheetDF<-sheetDF[order(sheetDF$arrayRow,sheetDF$arrayColumn),]
#Add a check on annotations from the gal file matching annotations from values
}#End readReplicates
readReplicatesAndGal<-function(dataFile){
#read a datafile with gal annotations
#Read the excel spreadsheet into a dataframe
values<-read.xls(dataFile, verbose=FALSE, stringsAsFactors=FALSE)
#Change Block to Grid
names(values)[which(names(values) %in% "Block")]="Grid"
#Add plate and replicate columns
values$plate=1
#Add a check on annotations from the gal file matching annotations from values
return(values)
}#End readReplicatesAndGal
simplifyNames<-function(df,pattern="Background|Net|Raw"){
#Find columns with a key name and simplify it to it's type and wavelength
colNames<-names(df)
temp<-lapply(colNames,FUN=function(colName,pattern,df){
m<-regexpr(pattern,colName)
if(m[[1]]!=-1){
#Found a column with one of the types of data in its name
dt<-regmatches(colName,m)
m<-regexpr(pattern="488|532|635|DAPI|594|647",colName)
if(m[[1]]!=-1){
#Found a column that has the right type of data and a wavelength
wl<-regmatches(colName,m)
#add a channel number prefix
######ch<-switch(EXPR=wl,"488"="Ch1","532"="Ch2","635"="Ch3","DAPI"="Ch1","594"="Ch2","647"="Ch3")
#create the new name for the column/channel
#####colName<-paste0(ch,dt,wl)
colName<-paste0(dt,wl)
}
}
return(colName)
},pattern=pattern)
names(df)<-temp
return(df)
}#End simplifyNames
makeNumeric<-function(df,pattern="Ch|Grid|Net|Background|Raw|X"){
#Find columns with a channel name
columns<-grep(pattern,names(df))
#Change each channel values to numeric
for(c in columns){
df[c]<-unlist(lapply(X=df[c],FUN=as.numeric))
}
return(df)
}#End makeNumeric
addControls<-function(df,pos=NULL,neg=NULL,empty=NULL){
#add a configuration column to the dataframe
#Wells are samples if they are not empty or a control
df$controlStatus<-rep('sample', dim(df)[1])
#Add any negative controls into this list
selectNeg<-df$name %in% c('Negative', neg)
selectNeg<-df$Name %in% c('Negative', neg)
df$controlStatus[selectNeg]<-'neg'
#Add any positive controls into this list
selectPos<-df$name %in% c('Positive',pos)
selectPos<-df$Name %in% c('Positive',pos)
df$controlStatus[selectPos]<-'pos'
selectEmpty<-df$name %in% c('Empty',empty)
selectEmpty<-df$Name %in% c('Empty',empty)
df$controlStatus[selectEmpty]<-'empty'
return(df)
}#End addControls
filterOnCellCount<-function(df,countCh,thresh=0){
#Replace the data values in all channels if the cell counts are below thresh or above the 99.9th percentile
df<-Data(df)
for (r in 1:dim(df)[2]){#go through all replicates
for(row in 1:dim(df)[1]){#search in all rows of ch 1 of this replicate for low values
#Force all low readings to NA values
if(df[row,r,countCh]<=thresh |
df[row,r,countCh] > quantile(df[,r,countCh],probs=.999, na.rm=TRUE)) df[row,r,]=NA
}
}
#Put the to cell count channel back into the cellHTS2 object
Data(x)<-df
return(x)
}#End filterLowCellCount
chtsAnnotate<-function(annDF,x){
#annotatae the x object with the annDF dataframe
tempFile<-tempfile()
write.table(annDF,file=tempFile, quote=FALSE, sep="\t")
x<-annotate(x, geneIDFile=tempFile)
return(x)
}#End chtsAnnotate
chtsNormalizePinGrids<-function(x,pinGridNorm='none'){
#Normalize all replicates and channels in x usingthe pinGridNorm method
if(pinGridNorm=="none") {
xpt<-x
} else {
xpt<-pinTipNormalize(x,method=pinGridNorm)
}
}#End chtsNormalizePinGrids
chtsNormalizePlates<-function(x,normMethod='none',logged=FALSE,refChannel=NULL,...){
#Wrapper to add reference channel method to normalize plates
if (normMethod!='none'){
if (normMethod=='refChannel') x<-refChannelNormalize(x, refChannel=refChannel,...)
else {
if(logged){
x<- suppressWarnings(normalizePlates(x,scale="additive", method=normMethod, ...))
}else {
x <- suppressWarnings(normalizePlates(x,scale="multiplicative", method=normMethod,...))
}
}
}
return(x)
}#End chtsNormalizePlates
trimSummaryStats<-function(df){
#Remove the summary statistics from the end of ProArray dataframe
if(length(which(df$X=="Mean value"))!=0){
tmp<-df[1:(which(df$X=="Mean value")-1),]
} else {
tmp<-df
}
return(tmp)
}#End trimSummaryStats
getAnnData<-function(x){
#extract the data, plate, grid(Grid), row and column from a cellHTS2 object and return it in a dataframe
#There is one column for each channel. Replicate values denoted by the replicate column
nrChannels<-dim(Data(x))['Channels']
nrReplicates<-dim(Data(x))['Samples']
channels<-data.frame(Data(x))
annotations<-fData(x)
#Find the replicate 1 channels by their names starting with 'X1.'
pattern<-'X1.'
repChannels <-grep(pattern,names(channels))
#Create a dataframe with the replicate one channels, the annotations and replicate =1
xout<-cbind(channels[repChannels],annotations,replicate=1)
#Strip off the X1. from the channel names
names(xout)[1:nrChannels]<-gsub("X[0-9].", "",names(channels))[repChannels]
#Repeat the same process for the rest of the replicates, and add the to the bottom of the dataframe
if (nrReplicates>1){
for(r in 2:nrReplicates){
pattern<-paste('X',r,sep="")
repChannels <-grep(pattern,names(channels))
xrep<-cbind(channels[repChannels],annotations,replicate=r)
names(xrep)[1:nrChannels]<-gsub("X[0-9].", "",names(channels))[repChannels]
xout<-rbind(xout,xrep)
}
}
xout$controlStatus<-factor(xout$controlStatus, levels=c('sample','pos','neg','other'), order=TRUE)
return(xout)
}
chtsPinBoxPlots<-function(x, main="",col='grey',...){
#Create boxplots of the pin tip values from a cellHTS2 object
xdf<-getAnnData(x)
#Read the number of wells in each plate from the object
nrWell <- prod(pdim(x))
#Read the number of plates in the screen from the object
nrPlates <- max(plate(x))
#Compute the number of Grids from the objects position data
nrGrids<-max(xdf$Grid)
#Read the number of channels from the object
nrChannels<-dim(Data(x))[3]
#Read the number of replicates
nrReplicates<-dim(Data(x))['Samples']
for(p in 1:nrPlates) {
for(r in 1:nrReplicates){
for(ch in 1:nrChannels){
boxplot(xdf[,ch][xdf$replicate==r]~xdf$Grid[xdf$replicate==r],data=xdf, groups=xdf$controlStatus,
col=col, aspect = "xy",
main=main,
#paste(main,'\n','Replicate', r, 'Channel',ch, 'Layout',p),
xlab='Pin Grid',ylab='Channel Value',...)
}
}
}
}
chtsPinQQs<-function(x,main=""){
#Create qqplots of the pin grids of a cellHTS2 object with different colors for the sample types.
library(lattice)
xdf<-getAnnData(x)
#Read the number of wells in each plate from the object
nrWell <- prod(pdim(x))
#Read the number of plates in the screen from the object
nrPlates <- max(plate(x))
#Compute the number of Grids from the objects position data
nrGrids<-max(xdf$Grid)
#Read the number of channels from the object
nrChannels<-dim(Data(x))[3]
for(p in 1:nrPlates) {
#Plates are not implemented yet(?)
for(r in 1:nrReplicates){
for(ch in 1:nrChannels){
plotqq<-qqmath(~ xdf[,ch][xdf$replicate==r] | Grid[xdf$replicate==r] , data=xdf, groups=controlStatus,
aspect = "1",pch=c(21,21,21),col=c('grey','black','black','black'),
fill=c('black','red','blue','magenta2'),
main=paste(main,'Pin Tip','\n',
'Replicate',r,'Channel',ch,'Plate',p), strip=strip.custom(strip.names = c(FALSE, FALSE)),
ylab='intensity',
prepanel = function(x, ...) {
list(xlim = range(qnorm(ppoints(length(x)))))
},layout=c(4,12),
panel = function(x, ...) {
qx <- qnorm(ppoints(length(x)))[rank(x)]
panel.xyplot(x = qx, y = x,..., aspect="1")
panel.abline(coef(lm(x~qx)), col='red', aspect="1")
})
print(plotqq)
}
}
}
}
chtsPlateQQs<-function(x,main=""){
#Create qqplots of the pin grids of a cellHTS2 object with different colors for the sample types.
library(lattice)
xdf<-getAnnData(x)
#Read the number of wells in each plate from the object
nrWell <- prod(pdim(x))
#Read the number of plates in the screen from the object
nrPlates <- max(plate(x))
#Compute the number of Grids from the objects position data
nrGrids<-max(xdf$Grid)
#Read the number of channels from the object
nrChannels<-dim(Data(x))[3]
for(p in 1:nrPlates) {
#Plates are not implemented yet(?)
for(r in 1:nrReplicates){
for(ch in 1:nrChannels){
plotqq<-qqmath(~ xdf[,ch][xdf$replicate==r] , data=xdf, groups=controlStatus,
aspect = "xy",pch=c(21,21,21),col=c('black','black','black','black'),
fill=c('black','red','blue','magenta2'),
main=paste(main,'Plate Values', '\n',
'Replicate ',r,'Channel',ch,'Plate',p), strip=strip.custom(strip.names = c(FALSE, FALSE)),
ylab='intensity',
prepanel = function(x, ...) {
list(xlim = range(qnorm(ppoints(length(x)))))
},layout=c(1,1),
panel = function(x, ...) {
qx <- qnorm(ppoints(length(x)))[rank(x)]
panel.xyplot(x = qx, y = x,...)
panel.abline(coef(lm(x~qx)), col='red')
})
print(plotqq)
}
}
}
}
chtsPinHists<-function(x,main=""){
#Create histograms of the pin grids of a cellHTS2 object
#TODO add a rug with different colors for the sample types
library(lattice)
xdf<-getAnnData(x)
#Read the number of wells in each plate from the object
nrWell <- prod(pdim(x))
#Read the number of plates in the screen from the object
nrPlates <- max(plate(x))
#Compute the number of Grids from the objects position data
nrGrids<-max(xdf$Grid)
#Read the number of channels from the object
nrChannels<-dim(Data(x))[3]
for(p in 1:nrPlates) {
#Plates are not implemented yet(?)
for(r in 1:nrReplicates){
for(ch in 1:nrChannels){
rawhist<-histogram(~xdf[,ch][xdf$replicate==r] | Grid[xdf$replicate==r], data=xdf,
groups=controlStatus,
layout = c(4,2),
xlab = "Intensity",
main=paste(main,'Pin Tip Values','\n',
'Replicate', r, 'Channel',ch,'Plate',p),
panel = function(x, ...) {
panel.histogram(x, col='grey', ...)
panel.rug(x,col='black', ...)
} )
print(rawhist)
}
}
}
}
chtsPlateHists<-function(x,main=""){
#Create histograms of the pin grids of a cellHTS2 object
#TODO add a rug with different colors for the sample types
library(lattice)
xdf<-getAnnData(x)
#Read the number of wells in each plate from the object
nrWell <- prod(pdim(x))
#Read the number of plates in the screen from the object
nrPlates <- max(plate(x))
#Compute the number of Grids from the objects position data
nrGrids<-max(xdf$Grid)
#Read the number of channels from the object
nrChannels<-dim(Data(x))[3]
for(p in 1:nrPlates) {
#Plates are not implemented yet(?)
for(r in 1:nrReplicates){
for(ch in 1:nrChannels){
rawhist<-histogram(~xdf[,ch][xdf$replicate==r] , data=xdf,
groups=controlStatus,
layout = c(1,1),
xlab = "Intensity",
main=paste(main,'Plate Values','\n',
'Replicate', r, 'Channel',ch,'Plate',p),
panel = function(x, ...) {
panel.histogram(x, col='grey', ...)
panel.rug(x,col='black', ...)
} )
print(rawhist)
}
}
}
}
chtsPinXY<-function(x, main=""){
#Create scatterplots of the pin grids of a cellHTS2 object
library(lattice)
#Create scatterplots of the raw pin grid values
xdf<-getAnnData(x)
#Read the number of wells in each plate from the object
nrWell <- prod(pdim(x))
#Read the number of plates in the screen from the object
nrPlates <- max(plate(x))
#Compute the number of Grids from the objects position data
nrGrids<-max(xdf$Grid)
#Read the number of channels from the object
nrChannels<-dim(Data(x))[3]
GridRows=max(xdf$Row)
GridCols<-max(xdf$Column)
#Assume there are 4 Grid columns in a subarray
GridGridCols=4
#First create a print order array
temp<-pseudoPrintOrder(blockCols=GridCols, blockRows=GridRows,
blockGridRows=nrGrids/GridGridCols, blockGridCols=GridGridCols)
for(p in 1:nrPlates) {
for(r in 1:nrReplicates){
for(ch in 1:nrChannels){
XY<-xyplot(xdf[,ch][xdf$replicate==r] ~ temp | Grid[xdf$replicate==r], data=xdf,
groups=controlStatus,pch=c(21,21,21,21),col=c('grey','red','blue','magenta2'),
fill=c('black','red','blue','magenta2'),layout = c(2,2),
xlab = "Order",
ylab='Intensity',
main=paste(main,'Pin Tip Values','\n',
'Replicate', r, 'Channel',ch,'Plate',p),
panel = function(x, y,...) {
panel.xyplot(x, y,...)
panel.abline(median(y),0, col='blue')
panel.loess(x,y, col='red')
})
print(XY)
}
}
}
}
chtsPlateXY<-function(x, main=""){
#Create scatterplots of the pin grids of a cellHTS2 object
library(lattice)
#Create scatterplots of the raw pin grid values
xdf<-getAnnData(x)
#Read the number of wells in each plate from the object
nrWell <- prod(pdim(x))
#Read the number of plates in the screen from the object
nrPlates <- max(plate(x))
#Compute the number of Grids from the objects position data
nrGrids<-max(xdf$Grid)
#Read the number of channels from the object
nrChannels<-dim(Data(x))[3]
GridRows=max(xdf$Row)
GridCols<-max(xdf$Column)
#Assume there are 4 Grid columns in a subarray
GridGridCols=4
#First create a print order array
temp<-pseudoPrintOrder(blockCols=GridCols, blockRows=GridRows,
blockGridRows=nrGrids/GridGridCols, blockGridCols=GridGridCols)
for(p in 1:nrPlates) {
for(r in 1:nrReplicates){
for(ch in 1:nrChannels){
XY<-xyplot(xdf[,ch][xdf$replicate==r] ~ temp , data=xdf,
groups=controlStatus,pch=c(21,21,21,21),col=c('grey','red','blue','magenta2'),
fill=c('black','red','blue','magenta2'),layout = c(1,1),
xlab = "Order",
ylab='Intensity',
main=paste(main,'Plate Values','\n',
'Replicate', r, 'Channel',ch,'Plate',p),
panel = function(x, y,...) {
panel.xyplot(x, y,...)
panel.abline(a=median(y),b=0, col='blue')
panel.loess(x,y, col='red')
})
print(XY)
}
}
}
}
chtsPlateByPinXY<-function(x, main=""){
#Create scatterplots of the pin grids of a cellHTS2 object
library(lattice)
#Create scatterplots of the raw pin grid values
xdf<-getAnnData(x)
for(p in 1:nrPlates) {
for(r in 1:nrReplicates){
for(ch in 1:nrChannels){
XY<-xyplot(xdf[,ch][xdf$replicate==r] ~ Grid , data=xdf,
groups=controlStatus,pch=c(21,21,21,21),col=c('grey','red','blue','magenta2'),
fill=c('black','red','blue','magenta2'),layout = c(1,1),
xlab = "Order",
ylab='Intensity',
main=paste(main,'Plate Values','\n',
'Replicate', r, 'Channel',ch,'Plate',p),
panel = function(x, y,...) {
panel.xyplot(x, y,...)
})
print(XY)
}
}
}
}
########## Rank Product of Replicates #########
chtsRankReplicates<-function(x,channels){
#Rank each replicate on a per channel basis, calculate the rank products
#across the replicates then store the p values of the rank products in the
#data slot, overwriting the normalized data
#Get the assay data
xd<-Data(x)
#Check the channel names and get the indices
chs<-which(channelNames(x) %in% channels)
if(length(chs)==0)stop("no channels matched the requested ones.")
#Get the number of replicates and check its more than 1
reps<-as.numeric(unlist(dimnames(xd)[2]))
if(length(reps)<=1)stop("Rank products require two or more replicates")
#Check the state of the object
if(!all(state(x)[c(1,2,4)]))
stop("The cellHTS2 object must be configured, normalized and annotated")
ranks<-matrix(nrow=dim(xd)[1],ncol=length(reps),
dimnames=list(NULL,paste0("ranks_r",colnames(xd))))
#order both replicates by their channel values
for(r in reps){
#The number one rank will have the highest value
ranks[,r]<-(1+dim(xd)[1])-rank(xd[,r,chs])
}
#Calculate the rank products
rp<-apply(X=ranks,MARGIN=1,FUN=prod)
#Get the rank product p value estimates using the gamma distribution
#This can be replaced with the exact p-values if needed
#rpp<-sapply(X=rp,FUN=righttailgamma,k=length(reps),n=dim(xd)[1])
k=length(reps)
n=dim(xd)[1]
#Debug Return the lower of the right tail or left tail p value
rpp<-pgamma(-log(rp/(n+1)^k),k,scale=1)
#Add the rpp scores into the object by replacing the normalized data
x<-summarizeReplicates(x)
Data(x)<-array(data=rpp,dim=c(length(rpp),1,1))
return(x)
}
########## End Rank Product of Replicates #######