R/pairwise_distances_timestamp.R

In sams25/rcombinator_old: Tools to Simulate the Evolution of Genetic Sequences

#' Gets pairwise distances from the sequence data at a particular timestamp
#' @param seq_data A data frame that contains a list of sequences
#' @param timestamp The timestamp at which our sequence data coressponds to
#' @return A data frame with the pairwise distances between the sequences for each time step
#' @importFrom magrittr "%>%"
#' @export
pairwise_distances_timestamp <- function(seq_data, timestamp, dist_method)
{
  seq_data <- seq_data[seq_data$time == timestamp, ]
  if(length(seq_data) == 0)
  {
    stop("Sequence data provided has no data for timestamp", timestamp)
  }
  num_sequences <- length(seq_data$sequences[[1]])
  num_seq_pairs <- (num_sequences * (num_sequences-1))/2
  size <- num_seq_pairs

  pair_data <- data.frame(time = rep(NA_integer_, size),
                          seq1 = rep(NA_integer_, size),
                          seq2 = rep(NA_integer_, size),
                          seq_dist = rep(NA, size))

  assign_row <- function(time_index, seq_data, num_seq_pairs)
  {
    if(dist_method == "ape")
    {
      new_rows <- dist_to_data_frame(stringdist_ape(seq_data$sequences[[1]]))
    }
    else if(dist_method == "levenshtein")
    {
      new_rows <- dist_to_data_frame(stringdist_levenshtein(seq_data$sequences[[1]]))
    }

    assign_row_inner <- function(seq_index, new_rows, time_index)
    {
      row_index <- seq_index
      new_row <- list(time_index,
                      new_rows$seq1[seq_index], new_rows$seq2[seq_index],
                      new_rows$dist_1_2[seq_index])
      pair_data[row_index, ] <<- new_row
    }
    purrr::map(1:num_seq_pairs, assign_row_inner, new_rows, time_index)
  }
  assign_row(timestamp, seq_data, num_seq_pairs)

  return(pair_data)
}