Benchmarking options at http://stackoverflow.com/questions/26049028/condense-merge-cells-in-a-table-in-r

SO26049028.R

## Required packages
library(data.table)
library(dplyr)
library(reshape2)
library(microbenchmark)

## Small sample `data.frame`
dat <- structure(list(Pos = c("Position1", "Position2", "Position3", 
"Position1", "Position3", "Position2"), M = c(34L, 45L, 89L, 
56L, 54L, 56L), P = c(56L, 23L, 78L, 45L, 35L, 89L)), .Names = c("Pos", 
"M", "P"), class = "data.frame", row.names = c(NA, -6L))

## Small sample `matrix`
m1 <- structure(c(34, 45, 89, 56, 54, 56, 56, 23, 78, 45, 35, 89), .Dim = c(6L, 
2L), .Dimnames = list(c("Position1", "Position2", "Position3", 
"Position1", "Position3", "Position2"), c("M", "P")))

## Built up to ~ 1M rows
dat <- do.call(rbind, replicate(16667, dat, FALSE))
m1 <- do.call(rbind, replicate(16667, m1, FALSE))


## rnso's double aggregate + merge
funRNSO <- function() {
  a1 = aggregate(M ~ Pos, dat, sum)
  a2 = aggregate(P ~ Pos, dat, sum)
  merge(a1, a2)
}

## my suggestion in comments
funRNSOb <- function() aggregate(. ~ Pos, dat, sum)

## Richard Scriven
funRS <- function() t(sapply(split(dat[-1], dat$Pos), colSums))

## Mine
funAM <- function() xtabs(Freq ~ Var1 + Var2, data.frame(as.table(m1)))

## Akrun - dplyr
funAdplyr <- function() {
  dat %>% 
    group_by(Pos) %>% 
    summarise_each(funs(sum=sum(., na.rm=TRUE)))
}

## Akrun - data.table
funAdt <- function() {
  as.data.table(dat)[, lapply(.SD, sum, na.rm=TRUE), by=Pos]
}

## Akrun - by
funAby <- function() {
  do.call(rbind, by(m1, list(rownames(m1)), colSums, na.rm=TRUE))
}

## Akrun - acast
funAacast <- function() {
  acast(melt(m1), Var1~Var2, value.var="value", sum, na.rm=TRUE)
}

system.time(funRNSO())
#    user  system elapsed 
#    3.20    0.00    3.22
system.time(funRNSOb())
#    user  system elapsed 
#    1.72    0.00    1.72

## ^^ Imagine if we had to aggregate more than 2 columns!

microbenchmark(tapply = funRS(), 
               xtabs = funAM(), 
               dplyr = funAdplyr(), 
               datatable = funAdt(), 
               by = funAby(), 
               acast = funAacast(), 
               times = 20)
# Unit: milliseconds
#       expr        min         lq    median        uq       max neval
#     tapply 155.003662 164.664556 175.49503 209.98393 275.68028    20
#      xtabs 463.060971 520.557225 580.92654 611.13589 720.41927    20
#      dplyr  17.555662  19.042972  21.12384  24.53657  29.02181    20
#  datatable   7.277477   8.253766  11.49202  18.26767  26.67410    20
#         by 232.103088 255.488470 288.51957 318.80311 378.95220    20
#      acast 159.206876 172.837925 213.22048 253.78435 331.10764    20