Vertically merge dataframes in R from a list

Question

[ellieuk]

Hi there,

I have some coverage stats and I want to merge them together. However, I want to merge them regardless of how many samples I have. Let's say in one batch I have 5 samples (A, B, C, D and E). Their corresponding files are (A.cov, B.cov, C.cov, D.cov and E.cov).

I first load these into R

library(tidyverse)
library(dplyr)

 #load in coverage.stats files
 coverage_stats = list.files(pattern="*.cov")
 for (i in 1:length(coverage_stats)) assign(coverage_stats[i], read.table(coverage_stats[i]))
 cov_stats_files = lapply(coverage_stats, read.table)

All is good here. This is the structure of the new dataframes (all the same) - this is A.cov:

structure(list(V1 = c("sample", "total_reads", "mapped_to_target_reads", 
"percentage", "mapped_to_target_reads_plus_150bp", "percentage"
), V2 = c("A", "56402158", "45018562", "79.82", "56664165", 
"100.46")), row.names = c(NA, 6L), class = "data.frame")

This is B.cov so you have another for good measure:

structure(list(V1 = c("sample", "total_reads", "mapped_to_target_reads", 
"percentage", "mapped_to_target_reads_plus_150bp", "percentage"
), V2 = c("B", "56402458", "45018555", "80.82", "5666416", 
"98")), row.names = c(NA, 6L), class = "data.frame")

I want to transform the tables into a nicer format:

transform_tables <- function(x) {
x %<>% t()
x <- as.data.frame(x)
x %<>% setNames(as.character(x[1,]))
x <- x[-1,] 
}

cov_stats_files <- lapply(cov_stats_files, transform_tables)

Now I have my tables in the format I want. I now want to bind all the tables vertically (like an rbind, but without explicitly giving the objects. I want to use the list cov_stats_files to do this (since each batch will have a different number of samples). This is where I'm stuck! I don't know how to iterate through the list and bind each dataframe together...

Would appreciate any help pls! E

Answer 1

[Ram]

You should be able to do that with Reduce(rbind, cov_stats_files), I think.

EDIT:

Use do.call(rbind, cov_stats_files) - it will be much faster as it uses rbind being a vectorized function to pass the entire list instead of passing the list's members two at a time (which is what Reduce does).

[ellieuk]

So easy! Thank you so much! Works beautifully! :-)

[Ram]

Courtesy rpolicastro - rbind is vectorized, so you can use do.call instead of Reduce as rbind doesn't need to be supplied arguments 2 at a time.

do.call(rbind, cov_stats_files)

will be faster than Reduce