Data Download & Setup

First, install the kassotis2020 package from GitHub and download the data from the EPA FTP.

devtools::install_github("daynefiler/kassotis2020")
library(kassotis2020)
library(data.table)
dataDir <- getwd() ## location for the downloaded zip files
downloadData(destDir = dataDir)

Extract and compile the PhI data:

ph1zip <- file.path(dataDir, "ToxCast_20110110.zip")
ph1fls <- c("ACEA", "Attagene", "BioSeek", "Cellumen", 
            "CellzDirect", "NCGC", "Novascreen")
ph1fmt <- "ToxCast_Phase_1_%s_20110110.txt"
ph1 <- lapply(sprintf(ph1fmt, ph1fls), freadZipFile, zipDir = ph1zip)
procPh1 <- function(x) {
  ## reformat CASRN to match phIII
  x[ , CODE := paste0("C", gsub("-", "", CASRN))] 
  x[ , c("GCID", "NAME", "CASRN") := NULL]
  setkey(x, CODE)
  x[]
}
ph1 <- Reduce(merge, lapply(ph1, procPh1))

ph1rnm <- ph1$CODE
## Make corrections for updated CASRN from Ph1 to Ph3
ph1rnm[ph1rnm == "C85007"] <- "CNOCAS44337"
ph1rnm[ph1rnm == "C51596113"] <- "CNOCAS34742"

ph1data <- data.matrix(ph1[ , -1])
rownames(ph1data) <- ph1rnm
ph1data <- -log10(ph1data/1e6)

The resulting data matrix, ph1data, is conveniently stored in the kassotis2020 package, and can be loaded into the environment using data(ph1data).

Extract and compile PhIII data:

ph3zip <- file.path(dataDir, "INVITRODB_V3_2_SUMMARY.zip")
chemData <- freadZipFile(ph3zip, "Chemical_Summary_190708.csv")
ph3fmt <- "%s_Matrix_190708.csv"
ph3fls <- c("modl_ga", "tested", "zscore", "hitc", "modl_tp")
ph3 <- lapply(sprintf(ph3fmt, ph3fls), freadZipFile, zipDir = ph3zip, drop = 1)
ph3 <- lapply(ph3, data.matrix)
names(ph3) <- ph3fls

## Transform the AC50 matrix
ph3$modl_ga <- 10^ph3$modl_ga
ph3$modl_ga[ph3$hitc == 0] <- 1e6
ph3$modl_ga[is.na(ph3$modl_ga) & ph3$tested == 1] <- 1e6
ph3$modl_ga <- -log10(ph3$modl_ga/1e6)

## Transform the efficacy matrix
ph3$modl_tp[ph3$hitc == 0] <- NA_real_
ph3$modl_tp <- apply(ph3$modl_tp, 2, scale)
colnames(ph3$modl_tp) <- paste0(colnames(ph3$modl_tp), "_emax")

## Transform the z-score matrix
ph3$zscore[ph3$zscore < 0] <- 0
ph3$zscore[ph3$hitc != 1] <- 0
ph3$zscore[is.na(ph3$zscore)] <- 0

ph3data <- cbind(ph3$modl_ga, ph3$modl_tp)
ph3zscr <- ph3$zscore

ph3rnm <- freadZipFile(ph3zip, sprintf(ph3fmt, ph3fls[1]), select = 1)[[1]]
rownames(ph3data) <- ph3rnm
rownames(ph3zscr) <- ph3rnm

Similarly, ph3data, ph3zscr, and chemData are available as data objects within the kassotis2020 package.

Dayne Filer & Chris Kassotis

2022-03-02