In anwarbio/bioassays: Summarising Multi Well Plate Cellular Assay

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>"
)

library(bioassays)

Libraries

The following packages will be useful. If they are not installed, please install them using install.packages().

library(tcltk)# for selecting the folder for analysis
library(dplyr)
library(ggplot2)# for plotting graphs
library(reshape2)
library(nplr)# for the standard curve fitting

Introduction

In a cell culture lab various cellular assays are performed. The package "bioassays" helps to analyse the results of these experiments performed in multiwell plates. The functions in this package can be used to summarise data from any multiwell plate, and by incorporating them in a loop several plates can be analyzed automatically. Two examples are shown in this article. All the csv files used for the examples are available in the exdata folder.

Example 1: Analysing result from a 96 well plate

Input data

To set up a folder as working directory

path1<-tk_choose.dir(getwd(), "Choose the folder for Analysis")
# A window will popup and ask you to select the folder containg the data  
setwd(path1)

Read files

Files can be read using the code example below. The files need to be in .csv format. In this example file names reflect the type of the assay used. For example "L_DIFF_P3_72HRS.csv" :L is assay code (Lactate assay), DIFF differentiation (type of cells used), p3: Plate 3 (each plate represent specific compounds tested), 72HRS 72hrs treatment with compound. This information will help to summarise the results.

filelist <-list("L_HEPG2_P3_72HRS.csv","L_HEPG2_P3_24HRS.csv") # list of files
fno<-1 # file number (in fileslist) that is going to be analyzed
result <- data.frame(stringsAsFactors= FALSE) ## An empty dataframe to dump result
zzz <- data.frame(stringsAsFactors= FALSE) ## An empty dataframe to dump result

To read the first file

filename<-extract_filename(filelist[fno])[1]
filename
nickname<-extract_filename(filelist[fno], split="_",end=".csv",remove="",sep="")[2]
nickname

rawdata<-read.csv(filename,stringsAsFactors = FALSE, strip.white = TRUE, 
                  na.strings = c("NA",""),header = TRUE,skip=1)
head(rawdata)

data(rawdata96)
rawdata<-rawdata96
head(rawdata)

Reading the metadata file

metadata<-read.csv("metafile.csv",stringsAsFactors = FALSE,strip.white = TRUE, 
                  na.strings = c("NA",""),header = TRUE)
head(metadata)

data(metafile96)
metadata<-metafile96
head(metadata)

Rearranging the data

96 well plates were used for the assay, so it is assumed that data is arrayed in rows A to G and columns 1 to 12 of a 96 well plate. The data2plateformat function is used to label them correctly.

rawdata<-data2plateformat(rawdata,platetype = 96)
head(rawdata)

To convert data into a data.frame.

The data2plateformat function uses the column and row names of the rawdata object to coerce it into a data.frame.

OD_df<- plate2df(rawdata)
head(OD_df)

Create a graphical overview

data<-matrix96(OD_df,"value",rm="TRUE")
heatplate(data,"Plate 1", size=5)

Filling metadata file using plate specific details

The example function given below determines the compound, concentration,type and dilution from the file name.

plate_info<-function(file,i){

  file<-file[1]
  plate<- extract_filename(file,split = "_",end = ".csv", remove = " ",
                           sep=" ")[5]

  if(plate == "P2"){
    compound<-"CyclosporinA"   # Concentration of cyclosporinA used for experiment
    concentration<-c(0,1,5,10,15,20,25,50) # Concentration of cyclosporinA used for experiment
    type<-c("S1","S2","S3","S4","S5","S6","S7","S8") # sample names of corresponding concentration
    dilution<-5
    plate_meta<-list(compound=compound,concentration=round(concentration,2),
                     type=type,dilution=dilution)
  }


  if(plate == "P3"){
    compound<-"Taxol"
    concentration<-c(0,0.0125,.025,.05,0.1,1,5,10) 
    type<-c("S1","S2","S3","S4","S5","S6","S7","S8")
    dilution<-5
    plate_meta<-list(compound=compound,concentration=round(concentration,2),
                     type=type, dilution=dilution)
  }


  if(plate =="p4"){
    compound<-c("Cisplatin")
    concentration<-c(0,0.5,2,4,8,16,64,"") 
    type<-c("S1","S2","S3","S4","S5","S6","S7","") 
    dilution <- 5
    plate_meta<-list(compound=compound,concentration=round(concentration,2),
                     type=type,dilution=dilution)
  }
  return(plate_meta)
}

plate_details<-plate_info(filelist,1)
plate_details

These 'plate details' can be used to add the metadata with the plate_metadata function.

metadata1<-plate_metadata(plate_details,metadata,mergeby="type")
head(metadata1)

For joining the metadata and platelayout dplyr's inner_join function can be used:

data_DF<- dplyr::inner_join(OD_df,metadata1,by=c("row","col","position"))
assign(paste("data",sep="_",nickname),data_DF) # create a copy of data with plate name
head(data_DF)

Sorting blank wells and reducing blanks

Blank values can be subtracted from the measurements with the reduceblank function.

data_DF<-reduceblank(data_DF, x_vector =c("All"),
                     blank_vector = c("Blank"), "value")
head(data_DF)
assign(paste("Blkmin",sep="_",nickname),data_DF) # create a copy of data with plate name

Plotting a standard curve

To filter standards

std<- dplyr::filter(data_DF, data_DF$id=="STD")  
std<- aggregate(std$blankminus ~ std$concentration, FUN = mean )
colnames (std) <-c("con", "OD")
head(std)

Calculations for standard curve : nonparametric logistic regression curve

fit1<-nplr::nplr(std$con,std$OD,npars=3,useLog = FALSE) 
#npars = 3 for 3 parametric regression curve

#for graph
x1 <- nplr::getX(fit1); y1 <- nplr::getY(fit1)
x2 <- nplr::getXcurve(fit1); y2 <- nplr::getYcurve(fit1)
plot(x1, y1, pch=15, cex=1, col="red", xlab="Concentration",
     ylab="Mean OD", main=paste("Standard Curve: ", nickname), cex.main=1)
lines(x2, y2, lwd=3, col="seagreen4")

To evaluate nonparametric logistic regression fitting

params<-nplr::getPar(fit1)$params
nplr::getGoodness(fit1)

To estimate the values based on logistic regression fitting

estimated_nplr<-estimate(data_DF,colname="blankminus",fitformula=fit1,method="nplr")
head(estimated_nplr)

Calculations for standard curve : linear regression curve

fit2<-stats::lm(formula = con ~ OD,data = std)
ggplot2::ggplot(std, ggplot2::aes(x=OD,y=con))+
ggplot2::ggtitle(paste("Std Curve:", nickname))+
ggplot2::geom_point(color="red",size=2)+
ggplot2::geom_line(data = ggplot2::fortify(fit2),ggplot2::aes(x=OD,y=.fitted),
          colour="seagreen4",size=1)+
ggplot2::theme_bw()

To evaluate the fit of the linear model

conpred<-estimate(std,colname="OD",fitformula=fit2,method="linear")
compare<-conpred[,c(1,3)]
corAccuracy<-cor(compare)[1,2]
corAccuracy ## Correlation accuracy of regression line
summary(fit2)

To estimate the values based on linear curve

estimated_lr<-estimate(data_DF,colname="blankminus",fitformula=fit2,
                       method="linear")
head(estimated_lr)

For multiply estimated by dilution

estimated_lr$estimated2 <- estimated_lr$estimated * estimated_lr$dilution
head(estimated_lr)

Summarise the data

For summarising the "estimated_lr" based on "id" and "type".

result<-dfsummary(estimated_lr,"estimated2",c("id","type"),
        c("STD","Blank"),"plate1", rm="FALSE",
        param=c(strict="FALSE",cutoff=40,n=12))

result

Statistical test

For the t test (S1 as control)

pval<-pvalue(result, control="S1", sigval=0.05)
head(pval)

\newpage

Example 2: Analysing result from a 384 well plate.

This example data contain result of dose response of few drugs (drug1, drug2, drug3, drug4) at three concentrations (C1,C2,C3) from two different cell lines (hepg2 and huh7)

Input data.

Reading the spectrophotometer readings from a csv file

rawdata2<-read.csv("384.csv",stringsAsFactors = FALSE,strip.white = TRUE, 
                  na.strings = c("NA",""),header = TRUE,skip=1)
dim(rawdata2)
head(rawdata2)

data(rawdata384)
rawdata2<-rawdata384
dim(rawdata2)
head(rawdata2)

Reading the metadata file

metadata2<-read.csv("metafile_384_plate.csv",stringsAsFactors = FALSE,
                    strip.white = TRUE, na.strings = c("NA",""), header = TRUE)
head(metadata2)

data(metafile384)
metadata2<-metafile384
head(metadata2)

Rearranging the data.

Renaming rows and columns of the rawdata2 object

rawdata2<-data2plateformat(rawdata2,platetype = 384)
head(rawdata2)

Converting the data into a data.frame.

OD_df2 <- plate2df(rawdata2)
head(OD_df2)

Create a graphical overview of the plate.

data2<-matrix96(OD_df2,"value",rm="TRUE")
heatplate(data2,"Plate 384", size=1.5)

Joining the metadata2 and OD_df2 objects

data_DF2<- dplyr::inner_join(OD_df2,metadata2,by=c("row","col","position"))
head(data_DF2)

Display the cell categories:

data3<-matrix96(data_DF2,"cell",rm="TRUE")
heatplate(data3,"Plate 384", size=2)

Discplay the compounds

data4<-matrix96(data_DF2,"compound",rm="TRUE")
heatplate(data4,"Plate 384", size=2)

Sorting blank wells and reducing blanks.

The data contain separate blanks wells for drug1, drug2, drug3 and drug4 (blank1, blank2,blank3 and blank 4 respectively) that can be used for background correction.

data_blk<-reduceblank(data_DF2, 
x_vector=c("drug1","drug2","drug3","drug4"),
blank_vector = c("blank1","blank2","blank3","blank4"), "value")
dim(data_blk)

head(data_blk)

For summarising the result in the order cell, compound,concentration,type and by omitting blanks.

result2<-dfsummary(data_blk,"blankminus",
        c("cell","compound","concentration","type"),
        c("blank1","blank2","blank3","blank4"),
        nickname="384well", 
        rm="FALSE",param=c(strict="FALSE",cutoff=40,n=12))
head (result2)
dim (result2)

The following example summarises the results in the following order: cell, compound and concentration, and by omitting blanks (all blanks are marked as "B" in concentration), drug 2 and huh7.

result3<-dfsummary(data_blk,"blankminus",
        c("cell","compound","concentration"),
        c("B","drug2","huh7"),
        nickname="", 
        rm="FALSE",param=c(strict="FALSE",cutoff=40,n=12))
head (result3)
dim (result3)

Statistical testing

The following command will perform a t-test:

pvalue<-pvalue(result3,"C3",sigval=0.05)
pvalue

anwarbio/bioassays documentation built on Oct. 9, 2020, 10:35 p.m.