Last updated: 2023-02-17

Checks: 5 1

Knit directory: combiDLBCL/analysis/

This reproducible R Markdown analysis was created with workflowr (version 1.7.0). The Checks tab describes the reproducibility checks that were applied when the results were created. The Past versions tab lists the development history.

Environment: empty

Great job! The global environment was empty. Objects defined in the global environment can affect the analysis in your R Markdown file in unknown ways. For reproduciblity it’s best to always run the code in an empty environment.

Seed: set.seed(20220425)

The command set.seed(20220425) was run prior to running the code in the R Markdown file. Setting a seed ensures that any results that rely on randomness, e.g. subsampling or permutations, are reproducible.

Session information: recorded

Great job! Recording the operating system, R version, and package versions is critical for reproducibility.

Cache: none

Nice! There were no cached chunks for this analysis, so you can be confident that you successfully produced the results during this run.

File paths: relative

Great job! Using relative paths to the files within your workflowr project makes it easier to run your code on other machines.

Repository version: no version control

Tracking code development and connecting the code version to the results is critical for reproducibility. To start using Git, open the Terminal and type git init in your project directory.

This project is not being versioned with Git. To obtain the full reproducibility benefits of using workflowr, please see ?wflow_start.

Load libraries and datasets

Read screen data

Prepare well annotation file

wellAnno <- readxl::read_excel("../data/Terzidou/Drug combo screening.xlsx", sheet = "RStudio Plate annotation") %>%
  mutate(rowID = paste0(Row,"0"), colID = sprintf("%02s",Column)) %>%
  mutate(wellID = paste0(rowID,colID),
         name = paste0(Drug_A,"_",Drug_B)) %>%
    mutate(ifEdge = rowID %in% c("A0","B0","O0","P0") | colID %in% c("01","02","23","24")) %>%
  select(wellID,  name, ifEdge, Drug_A, Drug_B, Drug_A.ConcStep, Drug_B.ConcStep, Drug_A.Conc, Drug_B.Conc, Drug_A.ConcUnit, Drug_B.ConcUnit)


write_tsv(wellAnno,"../output/AAscreen2022_wellAnno.tsv")

Prepare plate annotation file

plateFile <- list.files("../data/Terzidou/rawdata/", recursive = TRUE, pattern = "csv")
plateAnno <- tibble(fileName = plateFile) %>%
  filter(!str_detect(fileName, "E23_P23")) %>%
  mutate(plate = str_extract(fileName, ".*(?=/)"),
         cellLine = str_extract(fileName, "(?<=[0-9]_).*(?=_[:upper:])"),
         screenDate = as.Date(str_extract(fileName, "(?<=M_)[0-9]{8}(?=-)"),"%Y%m%d"),
         plateID = basename(fileName)) %>%
  mutate(cellLine = str_replace(cellLine,"WS-","WSU-")) %>% #fix some errors
  mutate(cellLine = str_replace(cellLine, "Su-","SU-"))
write_tsv(plateAnno, "../output/AAscreen2022_plateAnno.tsv")

Check the plate information

table(plateAnno$plate, plateAnno$cellLine)

                
                 Balm-3 DOHH-2 Farage HBL-1 HBL-1 Zenz HT K-422 Karpas-1106p
  Everolimus          1      1      1     1          0  1     1            1
  Ganetespib          2      1      1     1          0  1     1            1
  Ibrutinib           1      1      1     1          0  1     1            1
  Ixazomib            1      1      1     1          1  1     1            1
  MI-2                1      1      1     1          0  1     1            1
  MIK665              1      1      1     1          0  1     1            1
  NIKi (Janssen)      1      1      1     1          0  1     1            1
  Ribociclib          1      1      1     1          0  1     2            1
  Venetoclax          1      1      1     1          0  1     2            1
  Vincristine         1      1      2     1          0  1     1            1
                
                 OCI-LY-3 Pfeiffer RIVA SC-1 SU-DHL-2 SU-DHL-4 SU-DHL-4 Zenz
  Everolimus            1        1    1    1        1        1             1
  Ganetespib            1        1    1    1        1        1             0
  Ibrutinib             1        1    1    1        1        1             1
  Ixazomib              1        1    2    1        2        1             1
  MI-2                  1        1    2    1        1        1             0
  MIK665                1        1    2    1        1        1             0
  NIKi (Janssen)        1        1    2    1        1        1             0
  Ribociclib            1        1    1    1        1        1             0
  Venetoclax            2        1    1    1        1        1             0
  Vincristine           1        1    1    1        1        1             1
                
                 SU-DHL-5 SU-DHL-5 Zenz SU-DHL-6 SU-DHL-8 TMD-8 U-2932
  Everolimus            1             1        1        1     2      1
  Ganetespib            1             0        1        1     1      1
  Ibrutinib             1             1        1        1     1      1
  Ixazomib              1             1        1        1     2      1
  MI-2                  1             0        1        1     2      2
  MIK665                1             0        1        1     1      1
  NIKi (Janssen)        1             0        1        1     1      1
  Ribociclib            1             0        1        1     3      1
  Venetoclax            1             0        1        1     1      1
  Vincristine           1             1        1        2     1      2
                
                 U-2932-R1 U-2932-R2 U-2940 WSU-DLCL-2 WSU-FSCCL
  Everolimus             1         1      1          1         1
  Ganetespib             1         1      1          1         1
  Ibrutinib              2         1      1          1         1
  Ixazomib               1         2      1          1         1
  MI-2                   2         1      1          1         1
  MIK665                 1         1      1          1         1
  NIKi (Janssen)         1         2      1          1         1
  Ribociclib             1         1      1          1         1
  Venetoclax             1         2      1          1         1
  Vincristine            1         2      1          1         1

Read whole screen

screenData <- readScreen("../data/Terzidou/rawdata",
                         plateAnnotationFile = "../output/AAscreen2022_plateAnno.tsv",
                         wellAnnotationFile = "../output/AAscreen2022_wellAnno.tsv", 
                         rowRange = c(7,22), negWell = c("DMSO_DMSO"), posWell = c(),
                         colRange = 2, discardLayer = 2, normalization = FALSE, sep = ",")

Modifications of some annotations

Change the base drug name

screenData <- mutate(screenData, Drug_A = ifelse(Drug_A != "DMSO", plate, "DMSO"),
                     name = paste0(Drug_A, "_", Drug_B))

Remove wells with pipetting errors

errorPlate <- tibble(wellID = c("D002","D007","M021","M022","M023","M024","O023"),
                    plate = c("Venetoclax","Venetoclax","Ganetespib","Ganetespib","Ixazomib","Ixazomib","Everolimus"),
                    exclude = TRUE)
screenData <- screenData %>% left_join(errorPlate, by = c("wellID","plate")) %>%
  mutate(value = ifelse(is.na(exclude), value, NA)) %>%
  select(-exclude)

Remove values in all M022 wells (additional pipetting error)

screenData <- mutate(screenData, value = ifelse(wellID == "M022", NA, value))

Normalize

screenData <- normalizePlate(screenData, method = "negatives", discardLayer = 2)

Quality Control

Basic screen information

How many base drugs (Drug_A)

unique(screenData$Drug_A)

 [1] "DMSO"           "Everolimus"     "Ganetespib"     "Ibrutinib"     
 [5] "Ixazomib"       "MI-2"           "MIK665"         "NIKi (Janssen)"
 [9] "Ribociclib"     "Venetoclax"     "Vincristine"

What are the combi drugs (Drug_B)

unique(screenData$Drug_B)

 [1] "DMSO"                       "Apigenin"                  
 [3] "DCA"                        "MK886"                     
 [5] "Selisistat"                 "Thiomyristoyl"             
 [7] "3_TYP"                      "Hydroxychloroquine Sulfate"
 [9] "Bafilomycin A1"             "Methotrexate"              
[11] "Compound 3K"                "AZD3965"                   
[13] "V-9302"                     "FCCP"                      
[15] "IACS-010759"                "FX11"                      
[17] "Dorsomorphin"               "Epacadostat"               
[19] "BT2"                        "MK2206"                    
[21] "Etomoxir"                   "OSS_128167"                
[23] "Lonidamine"                 "C75"                       
[25] "6AN"                        "Clofarabine"               
[27] "Gamitrinib"                 "C646"                      
[29] "Fluvastatin"                "C2 ceramide"               
[31] "OT-82"                      "SHIN1"                     
[33] "DS18561882"                 "CB-839"                    
[35] "CPI-613"                    "Firsocostat"               
[37] "2-DG"                       "BAY-876"                   
[39] "CK37"                       "9-ING-41"                  
[41] "CAY10566"                   "ND-646"

Plot plate layout

screenData <- screenData %>% 
    mutate(type = case_when(
        Drug_A == "DMSO" & Drug_B == "DMSO" ~ "DMSO",
        Drug_A == "DMSO" & Drug_B != "DMSO" ~ "drug_only",
        Drug_A != "DMSO" & Drug_B == "DMSO" ~ "base_only",
        TRUE ~ "combine"
    ))

The two most outside layer is defined as edges.

Does every plate have the same layout?

plateIden <- group_by(screenData, plateID, wellID) %>%
    summarise(ifSame = length(unique(Drug_A))==1 &
                  length(unique(Drug_B))==1)
all(plateIden$ifSame)

[1] TRUE

Yes.

Plot drug type layout

plateLayout <- distinct(screenData, rowID, colID, Drug_A, Drug_B, type, Drug_B.ConcStep, Drug_A.ConcStep)

ggplot(plateLayout, aes(x=colID, y=rowID, fill = type)) +
    geom_tile() +
    theme_classic()

Plot Base drug layout

ggplot(plateLayout, aes(x=colID, y=rowID, fill = Drug_A)) +
    geom_tile() +
    theme_classic()

Plot combi drug layout

ggplot(plateLayout, aes(x=colID, y=rowID, fill = Drug_B)) +
    geom_tile() +
    theme_classic()

Plot base drug concentration layout

ggplot(plateLayout, aes(x=colID, y=rowID, fill = Drug_A.ConcStep)) +
    geom_tile() +
    theme_classic()

Plot combi drug concentration layour

ggplot(plateLayout, aes(x=colID, y=rowID, fill = Drug_B.ConcStep)) +
    geom_tile() +
    theme_classic()

Viability plate plot

pList <- lapply(unique(screenData$plateID), function(id){
    plotTab <- filter(screenData, plateID == id) 
    ggplot(plotTab, aes(x=colID, y=rowID, fill = normVal)) +
        geom_tile() + 
        scale_fill_gradient2(low = "blue", high = "red", mid = "white", midpoint = 1,limits=c(0,2)) +
        theme_classic() +
        ggtitle(id)
}) 
jyluMisc::makepdf(pList, "../docs/plate_plot_AA2022.pdf",ncol = 2, nrow = 3, width = 12, height = 12)

./plate_plot_AA2022.pdf

Check DMSO controls

Mean and Variance of internal controls

plotTab <- filter(screenData, !ifEdge, type == "DMSO")
ggplot(plotTab, aes(x=cellLine, y=value)) +
    ggbeeswarm::geom_quasirandom() +
    theme(axis.text.x = element_text(angle = 90, vjust=0.5, hjust = 1)) +
  facet_wrap(~plate)

Visualize edge effect

plotTab <- screenData %>% filter(type == "DMSO")
ggplot(plotTab, aes(x=cellLine, y=normVal, col = ifEdge)) +
    geom_boxplot() +
    theme(axis.text.x = element_text(angle = 90, vjust=0.5, hjust = 1))

Some degree of edge effect can be observed

Per row

plotTab <- screenData %>% filter(type == "DMSO")
ggplot(plotTab, aes(x=rowID, y=normVal)) +
    geom_boxplot() +
    theme(axis.text.x = element_text(angle = 90, vjust=0.5, hjust = 1))

Per column

plotTab <- screenData %>% filter(type == "DMSO")
ggplot(plotTab, aes(x=colID, y=normVal)) +
    geom_boxplot() +
    theme(axis.text.x = element_text(angle = 90, vjust=0.5, hjust = 1))

Looks like there’s strong edge effect in this screen.

Correct for edge effect

screenData <- DrugScreenExplorer::correctEdgeEffect(screenData)

After correction

plotTab <- screenData %>% filter(type == "DMSO")
ggplot(plotTab, aes(x=cellLine, y=normVal.cor, col = ifEdge)) +
    geom_boxplot() +
    theme(axis.text.x = element_text(angle = 90, vjust=0.5, hjust = 1))

Plate plot after edge correction

pList <- lapply(unique(screenData$plateID), function(id){
    plotTab <- filter(screenData, plateID == id) 
    ggplot(plotTab, aes(x=colID, y=rowID, fill = normVal.cor)) +
        geom_tile() + 
        scale_fill_gradient2(low = "blue", high = "red", mid = "white", midpoint = 1,limits=c(0,2)) +
        theme_classic() +
        ggtitle(id)
}) 
jyluMisc::makepdf(pList, "../docs/plate_plot_AA2022_edgeCor.pdf",ncol = 2, nrow = 3, width = 12, height = 12)

./plate_plot_AA2022_edgeCor.pdf

Visualize Combination design

drugTab <- distinct(screenData, Drug_A, Drug_B, Drug_A.Conc, Drug_B.Conc) %>%
  mutate(present = 1) %>%
  mutate(drugConcA = paste0(Drug_A,"_",Drug_A.Conc),
         drugConcB = paste0(Drug_B, "_", Drug_B.Conc))


countTab <- group_by(screenData, Drug_A, Drug_B, Drug_A.Conc, Drug_B.Conc, cellLine) %>%
  summarise(n=length(Drug_A)) %>%
  filter(! (Drug_B=="DMSO" & Drug_A == "DMSO"), n>1)

orderA <- arrange(drugTab, Drug_A, Drug_A.Conc)
orderB <- arrange(drugTab, Drug_B, Drug_B.Conc)

designMat <- drugTab %>% 
  select(drugConcA, drugConcB, present) %>%
  pivot_wider(names_from = drugConcB, values_from = present) %>%
  pivot_longer(-drugConcA) %>%
  mutate(name = factor(name, levels = unique(orderB$drugConcB)),
         drugConcA = factor(drugConcA, levels = unique(orderA$drugConcA)))

ggplot(designMat, aes(x=name, y=drugConcA, fill = value)) +
  geom_tile() +
  theme_minimal() +
  theme(axis.text.x = element_text(angle = 90, hjust = 1, vjust=0.5))

It follows diagonal design.

Visualize single drug effect

Dose response curves of base drugs

drugTab <- distinct(screenData, Drug_A, Drug_B, Drug_B.Conc) %>%
  filter(Drug_A != "DMSO", Drug_B.Conc==0) %>%
  mutate(id = paste0(Drug_A,"_",Drug_B,"_",Drug_B.Conc))

pList.single <- lapply(seq(nrow(drugTab)), function(i) {
  rec <- drugTab[i,]
  plotTab <- filter(screenData, 
                    Drug_A == rec$Drug_A,
                    Drug_B == rec$Drug_B,
                    Drug_B.Conc == rec$Drug_B.Conc) %>%
    group_by(Drug_A.Conc, cellLine) %>%
    summarise(normVal = mean(normVal, na.rm=TRUE))
  
  ggplot(plotTab, aes(x=Drug_A.Conc, y=normVal, group = cellLine, col = cellLine)) +
    geom_line() + geom_point() +
    scale_x_log10() + theme_bw() +
    coord_cartesian(ylim = c(0,1.5)) +
    ggtitle(rec$id) +
    theme(legend.position = "none")
})
names(pList.single) <- drugTab$id

jyluMisc::makepdf(pList.single, "../docs/DoseResponse_base_AA2022.pdf",2,3,width = 8, height = 8)

DoseResponse_base_AA2022.pdf

Edge effect corrected

drugTab <- distinct(screenData, Drug_A, Drug_B, Drug_B.Conc) %>%
  filter(Drug_A != "DMSO", Drug_B.Conc==0) %>%
  mutate(id = paste0(Drug_A,"_",Drug_B,"_",Drug_B.Conc))

pList.single <- lapply(seq(nrow(drugTab)), function(i) {
  rec <- drugTab[i,]
  plotTab <- filter(screenData, 
                    Drug_A == rec$Drug_A,
                    Drug_B == rec$Drug_B,
                    Drug_B.Conc == rec$Drug_B.Conc) %>%
    group_by(Drug_A.Conc, cellLine) %>%
    summarise(normVal.cor = mean(normVal.cor, na.rm=TRUE))
  
  ggplot(plotTab, aes(x=Drug_A.Conc, y=normVal.cor, group = cellLine, col = cellLine)) +
    geom_line() + geom_point() +
    scale_x_log10() + theme_bw() +
    coord_cartesian(ylim = c(0,1.5)) +
    ggtitle(rec$id) +
    theme(legend.position = "none")
})
names(pList.single) <- drugTab$id

jyluMisc::makepdf(pList.single, "../docs/DoseResponse_base_AA2022_cor.pdf",2,3,width = 8, height = 8)

DoseResponse_base_AA2022_cor.pdf

Dose response curves of combi drugs

drugTab <- distinct(screenData, Drug_A, Drug_B) %>%
  filter(Drug_A == "DMSO", Drug_B!= "DMSO")

pList.combi <- lapply(seq(nrow(drugTab)), function(i) {
  rec <- drugTab[i,]
  plotTab <- filter(screenData, 
                    Drug_A == rec$Drug_A,
                    Drug_B == rec$Drug_B) %>%
    group_by(Drug_B.Conc, cellLine) %>%
    summarise(normVal = mean(normVal, na.rm=TRUE))
  
  ggplot(plotTab, aes(x=Drug_B.Conc, y=normVal, group = cellLine, col = cellLine)) +
    geom_line() + geom_point() +
    scale_x_log10() + theme_bw() +
    coord_cartesian(ylim = c(0,1.5)) +
    ggtitle(paste0(rec$Drug_B)) +
    theme(legend.position = "none")
})
names(pList.combi) <- drugTab$Drug_B

jyluMisc::makepdf(pList.combi, "../docs/DoseResponse_single_AA2022.pdf",2,3,width = 8, height = 8)

DoseResponse_single_AA2022.pdf