R/data-blood_serum.R

In npaterno/data_hunter: Data set repository for OpenIntro textbooks

#' Acute Exercise, Prostate Cancer Cell Growth, and Immune System Cell Growth.
#'
#' Blood serum is a component of blood important to the immune system.
#' In this study, blood serum was collected from 10 men, both before
#' and immediately after exercise. Each of these serum samples, two from
#' each man, were then exposed to two types of cells. The first type,
#' LNCaP, is a type of prostate cancer cell. The second type, NIH3T3,
#' is an immune system cell. The growth of the two types of cells was
#' recorded after 48 hours.
#'
#' @name blood_serum
#' @docType data
#' @format A data frame with 40 rows and 4 variables.
#' \describe{
#'   \item{patient}{A number assigned to each patient.}
#'   \item{cell}{Type of cell, either "LNCaP" (prostate cancer cell)
#'     or "NIH3T3" (immune system cell).}
#'   \item{serum}{When the serum was collected, either "before" or "after" exercise.}
#'   \item{growth}{Cell proliferation, measured in arbitrary units.
#'     (Arbitrary units mean relative values are important but the
#'     absolute values are not.)}
#' }
#' @examples
#' library(ggplot2)
#' head(blood_serum)
#' table(blood_serum$patient)
#' qplot(
#'   ifelse(serum == "after", 1, 0),
#'   growth,
#'   data = blood_serum,
#'   geom = "line",
#'   color = patient
#' ) +
#'   facet_grid(cols = vars(cell)) +
#'   xlab("Before (0) and After (1)")
#' 
#' 
#' x <- subset(blood_serum, patient == 13 & cell == "NIH3T3")
#' blood_serum_change <- do.call(rbind.data.frame, by(
#'   blood_serum,
#'   data.frame(blood_serum$patient, blood_serum$cell),
#'   function(x) {
#'     x$change <- x$growth[x$serum == "after"] - x$growth[x$serum == "before"]
#'     x$growth <- NULL
#'     x$serum <- NULL
#'     return(x[1, ])
#'   }
#' ))
#' 
#' # t-test methodologies follow.
#' 
#' # First, look at the groups separately, which are similar to the analyses
#' # done in the paper.
#' bsc_cancer <- subset(blood_serum_change, cell == "LNCaP")
#' t.test(bsc_cancer$change)
#' bsc_immune <- subset(blood_serum_change, cell == "NIH3T3")
#' t.test(bsc_immune$change)
#' 
#' # In this next test, we treat the immune cells as the controls for the
#' # cancer cells from the corresponding individual, and the results are
#' # no longer statistically different from zero. This is effectively a
#' # "difference of differences" analysis.
#' # (Not 100% sure this comparison is valid, particularly if the units
#' # of growth vary from one cell type to the next.)
#' immune_to_cancer_map <- match(bsc_cancer$patient, bsc_immune$patient)
#' t.test(bsc_immune$change[immune_to_cancer_map] - bsc_cancer$change)
#' 
#' # Visualize the differences between the two types of cells for each
#' # of the 10 individuals.
#' qplot(
#'   ifelse(cell == "NIH3T3", 0, 1),
#'   change,
#'   data = blood_serum_change,
#'   geom = "line",
#'   color = patient
#' ) + xlab("Immune Cell (0) and Cancer Cell (1)")
#' # Another way to run the test using both cell types is with multiple
#' # regression. The result exactly matches the last t-test above.
#' m <- lm(
#'   change ~ cell + patient,
#'   data = blood_serum_change
#' )
#' summary(m)
#' @source [Rundqvist H, Augsten M, et al. 2013. Effect of Acute Exercise on Prostate Cancer Cell Growth. PLOS One 8(7):e67579](https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0067579).
#' @keywords datasets
#'
"blood_serum"