File size: 8,482 Bytes
1fa45cf
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
args <- commandArgs(trailingOnly=TRUE)
root <- if (length(args)) args[[1]] else "build/figures"
script_arg <- grep("^--file=", commandArgs(), value=TRUE)
script_dir <- dirname(normalizePath(sub("^--file=", "", script_arg[[1]])))
input <- if (length(args) >= 2) args[[2]] else "data"
dir.create(root, recursive=TRUE, showWarnings=FALSE)
png_device <- if (.Platform$OS.type == "windows") "windows" else "cairo"
read <- function(name) read.csv(file.path(input, name), stringsAsFactors=FALSE)
t <- read("tissue.csv")
te <- subset(read("tissue_effects.csv"), feature=="CIP2A")
g <- read("chronos.csv")
r <- subset(read("rnai.csv"), is.finite(dependency))
p <- read("pairs.csv")
p <- p[tolower(p$eligible_25)=="true",]
s <- read("cip2a_stability_grid.csv")
arms <- read("gse281740_cnv_arms.csv")
chan <- read("chan_confirmed_primary.csv")
ext <- subset(read("gse281740_cnv_effects.csv"), stratum=="all")
stopifnot(nrow(t)==122,nrow(g)==70,nrow(r)==11,nrow(subset(p,stratum=="all"))==4,nrow(s)==120)
pal <- c("#16736D","#A03951")
accent <- "#453C78"
emit <- function(name, panel, n=3) {
  for (fmt in c("pdf","png")) {
    file <- file.path(root,paste0(name,".",fmt))
    if (fmt=="pdf") pdf(file,width=3.35*n,height=3.25,useDingbats=FALSE)
    else png(file,width=3.35*n,height=3.25,units="in",res=300,type=png_device)
    par(mfrow=c(1,floor(n)),mar=c(4.5,4.5,2.1,1.2),mgp=c(2.7,.75,0),las=1,
        bty="l",cex=.83,font.main=2,col.axis="#222222")
    set.seed(45)
    panel()
    dev.off()
  }
}
ttl <- function(x) title(main=x,adj=0,line=.6,cex.main=1)
strip <- function(values, labs, y, zero=NULL) {
  plot(NA,xlim=c(.55,2.45),ylim=extendrange(unlist(values),f=.08),xaxt="n",xlab="",ylab=y)
  axis(1,1:2,labs)
  if (!is.null(zero)) abline(h=zero,lty=2,col="grey55")
  for (i in 1:2) {
    v <- values[[i]]
    points(i+runif(length(v),-.15,.15),v,pch=19,cex=.75,col=adjustcolor(pal[i],.7))
    segments(i-.22,median(v),i+.22,median(v),lwd=2)
  }
}
emit("FigureS6",function(){
  for(state in c("all","low_cycle")){
    z <- s[s$stratum==state,]
    z$setting <- match(z$seed,unique(s$seed))+5*(match(z$quantile,c(.95,.975,.99))-1)+15*(z$min_arms-2)
    z$eligible <- z$tumour_cells>=25 & z$epithelial_cells>=25
    cap <- if(state=="all") 45 else 1.5
    ramp <- colorRampPalette(c("#356999","#FFFFFF","#A03951"))(101)
    col <- ramp[1+round(50*(1+pmax(-1,pmin(1,z$delta_cpm/cap))))]
    col[!z$eligible] <- "#C6CDD3"
    y <- match(z$sample,c("ru1449a","ru1334"))
    stopifnot(nrow(z)==60,!anyDuplicated(data.frame(y,z$setting)))
    plot(NA,xlim=c(.5,30.5),ylim=c(.15,2.65),xaxt="n",yaxt="n",xlab="",ylab="",bty="n")
    rect(z$setting-.5,y-.26,z$setting+.5,y+.26,col=col,border="white")
    axis(2,1:2,c("RU1449A","RU1334"),las=2,cex.axis=.72)
    text(seq(3,28,5),2.39,rep(c("95%","97.5%","99%"),2),cex=.7)
    text(c(8,23),.57,c("2 arms","3 arms"),cex=.78)
    abline(v=15.5,col="grey60",lty=2)
    k <- seq(2,20,length.out=101)
    rect(k,.28,k+.19,.41,col=ramp,border=NA)
    text(c(2,11,20),.14,c(-cap,0,cap),cex=.65,xpd=NA)
    rect(23,.28,24.4,.41,col="#C6CDD3",border=NA)
    text(27,.34,"Excluded",cex=.64)
    mtext("Tumor minus epithelial (CPM)",side=1,line=2.15,cex=.76)
    ttl(if(state=="all") "A  All cells" else "B  Low-cycle marker stratum")
  }
  z <- subset(s,tumour_cells>=25 & epithelial_cells>=25 & matched_cells>=25)
  vals <- split(z$matched_positive,z$stratum)
  plot(NA,xlim=c(.55,2.45),ylim=c(0,1.05),xaxt="n",xlab="",ylab="Fraction of matching draws > 0")
  axis(1,1:2,c("All cells","Low-cycle markers"))
  for(i in 1:2) for(j in 1:2){
    q <- z[z$stratum==names(vals)[i]&z$sample==c("ru1334","ru1449a")[j],"matched_positive"]
    points(i+(j-1.5)*.25+runif(length(q),-.08,.08),q,pch=19,col=adjustcolor(pal[j],.68),cex=.75)
  }
  legend("bottomleft",c("RU1334","RU1449A"),pch=19,col=pal,bty="n",cex=.7)
  ttl("C  Repeated depth matching")
})
emit("FigureS1",function(){
  for(id in c("ru1334","ru1449a")){
    z <- subset(arms,sample==id)
    x <- xtabs(median_tumour~arm+reference,z)
    plot(x[,1],x[,2],pch=19,col=pal[1],xlab="Immune reference arm score",ylab="Stromal reference arm score")
    abline(h=0,v=0,col="grey60",lty=2)
    ttl(paste(if(id=="ru1334") "A" else "B",toupper(id)))
  }
},n=2)

emit("FigureS2",function(){
  classes <- c("malignant","nonmalignant_epithelial","immune","Fibroblast","Endothelial","other")
  cols <- c(pal[2],pal[1],"#527CB0","#C18A49","#745D91","#9A9A9A")
  make <- function(variable){
    z <- xtabs(reformulate(c("group","patient"),response=variable),chan)[classes,]
    sweep(z,2,colSums(z),"/")*100
  }
  for (v in c("cells","CIP2A")){
    barplot(make(v),col=cols,border=NA,ylim=c(0,100),names.arg=sub("HTA8_","",sort(unique(chan$patient))),
            ylab=if(v=="cells") "Fraction of cells (%)" else "Fraction of CIP2A UMI (%)")
    ttl(if(v=="cells") "A  Sample composition" else "B  Transcript contribution")
  }
  par(mar=rep(.5,4)); plot.new()
  legend("center",c("Tumor group","Nonmalignant epithelium","Immune","Fibroblast","Endothelial","Other"),
         fill=cols,border=NA,bty="n")
})
emit("FigureS3",function(){
  par(mar=c(.4,.6,1.2,.6))
  flow <- function(lines){
    plot.new(); plot.window(xlim=c(0,1),ylim=c(0,1))
    y <- seq(.89,.11,length.out=length(lines))
    h <- .082
    for(i in seq_along(lines)){
      if(i>1) arrows(.5,y[i-1]-h,.5,y[i]+h,length=.08,col="grey45")
      rect(.05,y[i]-h,.95,y[i]+h,col=if(i==length(lines)) "#DDEDEB" else "#F1F3F5",border="#8593A0")
      text(.5,y[i],lines[i],cex=.92)
    }
  }
  flow(c("Chan atlas\n9 patients with lung SCLC specimens",
         "4 patients with corroborating\nprimary lung labels",
         "Published cell annotations\nAt least 25 cells per group",
         "2 eligible patient pairs\n2 had only 4 or 5 epithelial cells"))
  flow(c("GSE281740\n19 specimens, 5 labeled primary lung",
         "3 overlapping patients removed\n2 independent candidates retained",
         "Marker and chromosome-arm assignment\nAt least 25 cells per group",
         "2 eligible patient pairs\nComputational cell labels"))
},n=2)
emit("FigureS4",function(){
  raw <- subset(read("gse281740_cnv_effects.csv"),stratum=="low_cycle_depth_matched")
  stopifnot(nrow(raw)==4)
  for(id in c("ru1334","ru1449a")){
    z <- raw[raw$sample==id,]
    z <- z[match(c("CNV_low_epithelial","CNV_supported_tumour"),z$group),]
    plot(NA,xlim=c(.65,2.35),ylim=c(-.18,2.45),xaxt="n",yaxt="n",xlab="",ylab="Observed CIP2A UMI")
    axis(1,1:2,c("Lung epithelium","Tumor group"))
    axis(2,0:2)
    segments(1:2,0,1:2,z$cip2a,col=pal,lwd=2)
    points(1:2,z$cip2a,pch=19,col=pal,cex=1.3)
    text(1:2,z$cip2a,labels=z$cip2a,pos=3)
    mtext(paste(z$cells,"cells per group")[1],side=3,line=-1.2,adj=.5,cex=.72)
    mtext(paste("Total UMI:",paste(format(z$umi,big.mark=",",trim=TRUE),collapse=" / ")),side=1,line=2.6,cex=.72)
    ttl(paste(if(id=="ru1334") "A" else "B",toupper(id)))
  }
},n=2)

emit("FigureS5",function(){
  par(mar=c(1.2,7.3,2.1,2.3))
  z <- r[order(r$dependency),]
  vals <- as.matrix(z[c("chronos","dependency")])
  ranges <- apply(vals,2,range)
  ramp <- colorRampPalette(c("#6C2469","#F1D6E7","#FAF8FA"))(101)
  plot(NA,xlim=c(.5,2.5),ylim=c(.3,12.3),xaxt="n",yaxt="n",xlab="",ylab="",bty="n")
  for(k in 1:2){
    y <- rev(seq_len(nrow(z)))+1
    pos <- (vals[,k]-ranges[1,k])/diff(ranges[,k])
    rect(k-.39,y-.44,k+.39,y+.44,col=ramp[1+round(pos*100)],border="white")
    text(k,y,sprintf("%.3f",vals[,k]),col=ifelse(pos<.26,"white","#262332"),cex=.84)
    text(k,.88,sprintf("%.2f to %.2f",ranges[1,k],ranges[2,k]),cex=.64)
  }
  text(c(1,2),12.76,c("CRISPR Chronos","RNAi DEMETER2"),xpd=NA,font=2,cex=.87)
  text(.56,y,z$model_name,adj=1,xpd=NA,cex=.78,
       font=ifelse(z$model_name %in% c("NCI-H211","NCI-H1836","DMS 53"),2,1))
  mtext("Darker cells indicate more negative effects within each platform",side=1,line=.05,cex=.7)
},n=1.75)

emit("FigureS7",function(){
  hi <- r$model_name %in% c("NCI-H211","DMS 53","NCI-H1836")
  plot(r$chronos,r$dependency,pch=19,col=ifelse(hi,pal[2],"grey55"),
       xlim=c(-.9,.2),ylim=c(-1,.16),xlab="CRISPR Chronos",ylab="RNAi DEMETER2")
  text(r$chronos[hi],r$dependency[hi],r$model_name[hi],pos=3,cex=.69)
  ttl("Matched screen effects")
},n=1.65)
source(file.path(script_dir,"plate_layouts.R"),encoding="UTF-8")