R/tabulate_shared_TCR_chains.R

In mjdufort/TCRtools: Helper functions for analyzing TCR sequence data from single-cell RNA-seq

#' Tabulate the shared TCRs chains in pairs of single-cell libraries
#'
#' Aggregate a set of TCR chain matches to generate a table with pairs of libraries and the number
#' of chains shared between the libraries.
#'
#' @param tcr_chain_matches a data frame containing the TCR chain matches. Should contain, at minimum, the identifiers for the two TCRs. Often the output of \code{match_TCR_chains}.
#' @param tcr1_col,tcr2_col numbers or names of the columns containing the identifiers
#' @importFrom stats aggregate na.omit
#' @importFrom utils read.table
#' @export
#' @return A data frame containing the two TCR identifiers and the number of chains shared between them.
#'
#' @usage
#' tabulate_shared_TCR_chains(
#'      tcr_chain_matches,
#'      tcr1_col="tcr1", tcr2_col="tcr2")
tabulate_shared_TCR_chains <-
  function(tcr_chain_matches,
           tcr1_col="tcr1", tcr2_col="tcr2") {
    checkmate::assert(
      checkmate::check_data_frame(tcr_chain_matches)
    )

    if (nrow(tcr_chain_matches) == 0) {
      tcr_chains_shared <-
        cbind(
          tcr_chain_matches[, c(tcr1_col, tcr2_col)],
          "num_shared_chains" = integer())
    } else {
      # convert numeric column identifiers to column names
      for (i in c("tcr1_col", "tcr2_col"))
        if (is.numeric(get(i))) assign(i, colnames(tcrs)[get(i)])

      # extract all the matches
      tcr_chains_shared <-
        aggregate(
          list(num_shared_chains=rep(1, nrow(tcr_chain_matches))),
          tcr_chain_matches[,c(tcr1_col, tcr2_col)], length)

      # make the order consistent, and drop the duplicates
      for (i in 1:nrow(tcr_chains_shared)) {
        tcr_chains_shared[i, c(tcr1_col, tcr2_col)] <-
          tcr_chains_shared[i, c(tcr1_col, tcr2_col)] %>%
          unlist() %>%
          sort()
      }
      tcr_chains_shared <- unique(tcr_chains_shared)
    }

    tcr_chains_shared
  }