Calculates the fraction of G+C bases of the input nucleic acid
sequence(s). It reads in nucleic acid sequences, sums the number of
'g' and 'c' bases and writes out the result as the fraction (in the
interval 0.0 to 1.0) to the total number of 'a', 'c', 'g' and 't' bases.
Global G+C content GC, G+C in the first position of the codon bases
GC1, G+C in the second position of the codon bases
GC2, and G+C in the third position of the codon bases
GC3 can be computed. All functions can take ambiguous bases
into account when requested.
GC(seq, forceToLower = TRUE, exact = FALSE, NA.GC = NA, oldGC = FALSE,
alphabet = s2c("acgtswmkryvhdb"))
GC1(seq, frame = 0, ...)
GC2(seq, frame = 0, ...)
GC3(seq, frame = 0, ...)
GCpos(seq, pos, frame = 0, ...)GC returns the fraction of G+C (in [0,1]) as a numeric vector of length one.
GCpos returns GC at position pos.
GC1, GC2, GC3 are wrappers for GCpos with the
argument pos set to 1, 2, and 3, respectively.
NA is returned when seq is NA.
NA.GC defaulting to NA is returned when the G+C content
can not be computed from data.
a nucleic acid sequence as a vector of single characters
for coding sequences, an integer (0, 1, 2) giving the frame
logical. if TRUE force sequence
characters in lower-case. Turn this to FALSE to save time
if your sequence is already in lower-case (cpu time is approximately
divided by 3 when turned off)
logical: if TRUE ambiguous bases are taken
into account when computing the G+C content (see details).
Turn this to FALSE to save time if your you can neglect
ambiguous bases in your sequence (cpu time is approximately
divided by 3 when turned off)
what should be returned when the GC is impossible to
compute from data, for instance with NNNNNNN. This behaviour could
be different when argument exact is TRUE, for instance
the G+C content of WWSS is NA by default, but is 0.5 when
exact is set to TRUE
arguments passed to the function GC
for coding sequences, the codon position (1, 2, 3) that should be taken into account to compute the G+C content
logical defaulting to FALSE: should the GC content computed
as in seqinR <= 1.0-6, that is as the sum of 'g' and 'c' bases divided by
the length of the sequence. As from seqinR >= 1.1-3, this argument is
deprecated and a warning is issued.
alphabet used. This allows you to choose ambiguous bases used during GC calculation.
D. Charif, L. Palmeira, J.R. Lobry
When exact is set to TRUE the G+C content is estimated
with ambiguous bases taken into account. Note that this is time expensive.
A first pass is made on non-ambiguous bases to estimate the probabilities
of the four bases in the sequence. They are then used to weight the
contributions of ambiguous bases to the G+C content. Let note nx
the total number of base 'x' in the sequence. For instance
suppose that there are nb bases 'b'. 'b' stands for "not a", that
is for 'c', 'g' or 't'. The contribution of 'b' bases to the GC base
count will be:
nb*(nc + ng)/(nc + ng + nt)
The contribution of 'b' bases to the AT base count will be:
nb*nt/(nc + ng + nt)
All ambiguous bases contributions to the AT and GC counts are weighted is similar way and then the G+C content is computed as ngc/(nat + ngc).
citation("seqinr").
The program codonW used here for comparison is available at https://codonw.sourceforge.net/.
mysequence <- s2c("agtctggggggccccttttaagtagatagatagctagtcgta")
GC(mysequence) # 0.4761905
GC1(mysequence) # 0.6428571
GC2(mysequence) # 0.3571429
GC3(mysequence) # 0.4285714
#
# With upper-case characters:
#
myUCsequence <- s2c("GGGGGGGGGA")
GC(myUCsequence) # 0.9
#
# With ambiguous bases:
#
GC(s2c("acgt")) # 0.5
GC(s2c("acgtssss")) # 0.5
GC(s2c("acgtssss"), exact = TRUE) # 0.75
#
# Missing data:
#
stopifnot(is.na(GC(s2c("NNNN"))))
stopifnot(is.na(GC(s2c("NNNN"), exact = TRUE)))
stopifnot(is.na(GC(s2c("WWSS"))))
stopifnot(GC(s2c("WWSS"), exact = TRUE) == 0.5)
#
# Coding sequences tests:
#
cdstest <- s2c("ATGATG")
stopifnot(GC3(cdstest) == 1)
stopifnot(GC2(cdstest) == 0)
stopifnot(GC1(cdstest) == 0)
#
# How to reproduce the results obtained with the C program codonW
# version 1.4.4 writen by John Peden. We use here the "input.dat"
# test file from codonW (there are no ambiguous base in these
# sequences).
#
inputdatfile <- system.file("sequences/input.dat", package = "seqinr")
input <- read.fasta(file = inputdatfile) # read the FASTA file
inputoutfile <- system.file("sequences/input.out", package = "seqinr")
input.res <- read.table(inputoutfile, header = TRUE) # read codonW result file
#
# remove stop codon before computing G+C content (as in codonW)
#
GC.codonW <- function(dnaseq, ...){
GC(dnaseq[seq_len(length(dnaseq) - 3)], ...)
}
input.gc <- sapply(input, GC.codonW, forceToLower = FALSE)
max(abs(input.gc - input.res$GC)) # 0.0004946237
plot(x = input.gc, y = input.res$GC, las = 1,
xlab = "Results with GC()", ylab = "Results from codonW",
main = "Comparison of G+C content results")
abline(c(0, 1), col = "red")
legend("topleft", inset = 0.01, legend = "y = x", lty = 1, col = "red")
if (FALSE) {
# Too long for routine check
# This is a benchmark to compare the effect of various parameter
# setting on computation time
n <- 10
from <-10^4
to <- 10^5
size <- seq(from = from, to = to, length = n)
res <- data.frame(matrix(NA, nrow = n, ncol = 5))
colnames(res) <- c("size", "FF", "FT", "TF", "TT")
res[, "size"] <- size
for(i in seq_len(n)){
myseq <- sample(x = s2c("acgtws"), size = size[i], replace = TRUE)
res[i, "FF"] <- system.time(GC(myseq, forceToLower = FALSE, exact = FALSE))[3]
res[i, "FT"] <- system.time(GC(myseq, forceToLower = FALSE, exact = TRUE))[3]
res[i, "TF"] <- system.time(GC(myseq, forceToLower = TRUE, exact = FALSE))[3]
res[i, "TT"] <- system.time(GC(myseq, forceToLower = TRUE, exact = TRUE))[3]
}
par(oma = c(0,0,2.5,0), mar = c(4,5,0,2) + 0.1, mfrow = c(2, 1))
plot(res$size, res$TT, las = 1,
xlab = "Sequence size [bp]",
ylim = c(0, max(res$TT)), xlim = c(0, max(res$size)), ylab = "")
title(ylab = "Observed time [s]", line = 4)
abline(lm(res$TT~res$size))
points(res$size, res$FT, col = "red")
abline(lm(res$FT~res$size), col = "red", lty = 3)
points(res$size, res$TF, pch = 2)
abline(lm(res$TF~res$size))
points(res$size, res$FF, pch = 2, col = "red")
abline(lm(res$FF~res$size), lty = 3, col = "red")
legend("topleft", inset = 0.01,
legend = c("forceToLower = TRUE", "forceToLower = FALSE"),
col = c("black", "red"), lty = c(1,3))
legend("bottomright", inset = 0.01, legend = c("exact = TRUE", "exact = FALSE"),
pch = c(1,2))
mincpu <- lm(res$FF~res$size)$coef[2]
barplot(
c(lm(res$FF~res$size)$coef[2]/mincpu,
lm(res$TF~res$size)$coef[2]/mincpu,
lm(res$FT~res$size)$coef[2]/mincpu,
lm(res$TT~res$size)$coef[2]/mincpu),
horiz = TRUE, xlab = "Increase of CPU time",
col = c("red", "black", "red", "black"),
names.arg = c("(F,F)", "(T,F)", "(F,T)", "(T,T)"), las = 1)
title(ylab = "forceToLower,exact", line = 4)
mtext("CPU time as function of options", outer = TRUE, line = 1, cex = 1.5)
}
Run the code above in your browser using DataLab