Studies scaling

library(ggplot2)
library(ggrepel)

Read Valentine’s DB

db <- read.delim('sdb_May2020.csv',sep=',')

db$Year <- as.numeric(substr(db$Date, 1, 4))
db$Reported.cells.total <- as.integer(gsub(',','',as.character(db$Reported.cells.total)))

time.scale

[1] 1420088400 1451624400

Select studies

# processors
tdf <- data.frame(year=c(1993,1997,2000,2002,2006,2008,2012,2014,2016,2017,2019),name=c('Pentium','AMD K6','Pentium III','Pentium IV Northwood','Core 2 Duo Conroe','Core i7','Core i7 Haswell','Xeon Hawell5','Xeon Broadwell','Epyc','Epyc ROME'),tc=c(3.1e6,8e6,2.1e7,5.5e7,291e6,731e6,1.4e9,5.5e9,7.2e9,19.2e9,39e9))

scf <- data.frame(year=c(2009,2010,2011,2013,2014,2015,2017,2019),
                  name=c('Surani','Surani','Linnarsson','West','STRT-C1','InDrops/Drop-seq','10x Chromium','sci-RNA-seq3'),
                  tc=c(2,15,85,301,3e3,44e3,1.3e6,2.06e6))

scf <- data.frame(do.call(rbind,list(
  c(2009,'Surani',2),
  c(2010,'Surani',15),
  c(2011.3,'STRT',85),
  c(2014.1,'MARS',4468),
  c(2015.1,'CytoSeq',15000),
  c(2015.38,'InDrops/Drop-seq',44e3),
  c(2017,'10x Chromium',1.3e6),
  c(2019,'sci-RNA-seq3',2.06e6)
)),stringsAsFactors=F)
colnames(scf) <- c('year','name','tc')
scf$year <- as.numeric(scf$year); scf$tc <- as.numeric(scf$tc)


tdf$type <- 'CPUs'
scf$type <- 'Landmark studies'
acf <- data.frame(year=db$time,name=db$Shorthand,tc=db$Reported.cells.total,type='all studies'); 

acf

cd <- do.call(rbind,list(tdf,scf,acf))
cd$type <- as.factor(cd$type)
cd$name <- as.character(cd$name)
cd$name[cd$type=='all studies'] <- '';

p <- ggplot(cd,aes(x=year,y=log10(tc),color=type,label=name)) + 
  geom_point(data=cd[cd$type=='all studies',],aes(x=year,y=log10(tc)),color=adjustcolor('red',alpha=0.2),size=0.2) +
  geom_point(size=2) + 
  coord_cartesian(ylim = c(0, 13.5)) +
  geom_hline(yintercept = log10(3.73*1e13),lty=2,col='darkgreen',size=1.5)+
  annotate('text',label='number of cells in a human body',x=Inf,y=12.8,vjust=0,hjust=1.3,size=5,color='darkgreen')+
  theme_bw() + guides(color='none') + 
  scale_color_manual(values=c(adjustcolor('white',alpha=0.01),"black", "red")) +
  geom_text_repel(box.padding = 0.55,size=4) +
  geom_smooth(method = "lm", se = FALSE,linetype=2,size=0.5,fullrange=TRUE) +
  xlab("Date")+ylab("log10[ count ]") 
  
p

pdf(file='studies.pdf',width=5,height=5); print(p); dev.off();

null device 
          1