R/plot_coloc.R

In otargen: Access Open Target Genetics

#' Scatter plot of colocalisations for gene and reported traits.
#'
#' Generates a scatter plot using the results from \code{colocalisationsForGene()}
#' function as an input. The reported trait in each study are shown on the
#' \emph{x-axis} and plotted against their corresponding \code{-log2(H4/H3)}
#' values on the \emph{y-axis}, indicating the evidence of colocalisation between
#' the molecular QTLs reported in each study and the explored gene. The molecular
#' QTLs are mapped to the colors of the points. If the results of
#' \code{colocalisationsForGene()} includes the data for multiple genes, they will
#' be plotted in separate panels.
#'
#' @param data Data Frame: result of colocalisationsForGene function in data
#' frame format, contacting the phewas information for a variant id
#' @param biobank Logical: \code{TRUE} and \code{FALSE} variable, with the default value of
#' FALSE which will keep the data that are from UKbioBank beside the published GWAS
#' data. In case, this parameter is set to \code{TRUE}, only UKbioBank data will be kept
#' which has not been published.
#'
#' @return A horizontal bar plot for colocalisation of information.
#'
#' @examples
#' \dontrun{
#' plot_out <- colocalisationsForGene(genes = "ENSG00000169174") %>%
#'           plot_coloc(biobank = FALSE)
#' plot_out <- colocalisationsForGene(genes = "PCSK9") %>%
#'           plot_coloc(biobank = TRUE)
#' }
#' @import dplyr
#' @import ggplot2
#' @importFrom magrittr %>%
#' @export
#'
#'
plot_coloc <- function(data, biobank = FALSE) {
  dt0 <- data
  dt0$study.traitCategory <- base::tolower(dt0$Trait_reported)
  dt1 <- dt0 %>%
    dplyr::mutate(Trait_reported_trimmed = stringr::str_replace_all(Trait_reported, pattern = "[:punct:]|[:symbol:]", replacement = "")) %>%
    dplyr::mutate(Trait_reported_trimmed = stringr::str_trunc(Trait_reported_trimmed, width = 35, side = "right"))

  dt2 <- dt1[!duplicated(dt1[ , c("Study","Lead_variant","Molecular_trait")]),]


  dt3 <- dt2 %>% dplyr::group_by(Molecular_trait,Lead_variant) %>%
    dplyr::arrange(desc(dt2$`log2(H4/H3)`)) %>% dplyr::top_n(1) %>%
    dplyr::select(Trait_reported_trimmed, Molecular_trait,Lead_variant,
                  Study,Tissue, Source,`log2(H4/H3)`) %>% dplyr::ungroup() %>%
    dplyr::filter(`log2(H4/H3)` > 7)


  # source <- match.arg(source)
  # type <- match.arg(type)

  if (biobank == TRUE) {
    dt4 <- dt3 %>% dplyr::filter(grepl(pattern = "^GCST.*", Study))
  } else {
    dt4 <- dt3

  }

 p <- dt4 %>% ggplot2::ggplot(ggplot2::aes(x= reorder(Trait_reported_trimmed, -`log2(H4/H3)`),
                                            `log2(H4/H3)`, fill = Source)) +
    ggplot2::geom_bar(stat = "identity", position = "dodge") +
    ggplot2::coord_flip () +
    ggplot2::facet_wrap(.~Molecular_trait) +

    ggplot2::geom_label(ggplot2::aes(
     label = Lead_variant,
     fill = Source),
   color = "white", fontface = "bold",  label.size = 0.1) +
    ggplot2::xlab("") +
    ggplot2::theme(
      panel.grid.major.y = ggplot2::element_blank(),
      panel.grid.minor.y = ggplot2::element_blank()
    )
  return(p)
}