R/get_summary_stats_timestep.R

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

#' Prints summary statistics for distances between sequences for a single timestep
#' @param seq_data The data returned by a simulation for a single timestep
#' @param timestep What this timestep is
#' @return A data frama of summary statistics for that time
#' @importFrom magrittr "%>%"
#' @export

get_summary_stats_timestep <- function(seq_data, timestep)
{
  summary_table <- data.frame(time = rep(NA_integer_, 1),
                              num_seq = rep(NA_integer_, 1),
                              mean_dist = rep(NA_real_, 1),
                              median_dist = rep(NA_real_, 1),
                              num_na = rep(NA_integer_, 1))
  summary_table$quantiles_dist <- list(rep(NA_real_, 4))
  
  num_seq <- max(c(seq_data$seq1, seq_data$seq2), na.rm = TRUE)
  mean_dist <- mean(seq_data$seq_dist, na.rm = TRUE)
  median_dist <- median(seq_data$seq_dist, na.rm = TRUE)
  quantiles_dist <- stats::quantile(seq_data$seq_dist, na.rm = TRUE)
  num_na <- sum(is.na(seq_data$seq_dist))

  summary_table[1, "time"] <- timestep
  summary_table[1, "num_seq"] <- num_seq
  summary_table[1, "mean_dist"] <- mean_dist
  summary_table[1, "median_dist"] <- median_dist
  summary_table[1, "num_na"] <- num_na
  summary_table$quantiles_dist[1] <- list(quantiles_dist)
  return(summary_table)
}