Hi,

I am trying to plot protein position weighted matrices in seqLogo and DiffLogo and seem to be have a similar problem in which the correct letter for the amino acid is not displayed properly in the plot.  It works fine for DNA sequences, but not proteins. 

Cheers, Jeff

library(Biostrings)
library(seqLogo)

#Logo of DNA alignment
sequences <- c("ATCCG","TACCG","AACGC")
pwm1 <- consensusMatrix(sequences, as.prob = TRUE)
pwm1
seqLogo(pwm = pwm1)

#Logo of protein alignment
sequences <- c("SCPCA", "PCPTQ", "ECPCQ")
pwm2 <- consensusMatrix(sequences, as.prob = TRUE)
pwm2
seqLogo(pwm=pwm2, alphabet=ASN)

Hi,

Package motifStack can plot for proteins. I am sorry for annoying you when I mention motifStack here.

Please have a try:

library(Biostrings)
library(seqLogo)
library(motifStack)
#Logo of DNA alignment
sequences <- c("ATCCG","TACCG","AACGC")
pwm1 <- consensusMatrix(sequences, as.prob = TRUE)
pwm1
seqLogo(pwm = pwm1)

#Logo of protein alignment
sequences <- c("SCPCA", "PCPTQ", "ECPCQ")
pwm2 <- consensusMatrix(sequences, as.prob = TRUE)
pwm2
#seqLogo(pwm=pwm2, alphabet=ASN)
pfm2 <- new("pfm", mat=pwm2, name="pwm2", alphabet="AA")
plot(pfm2)

Thanks!  This above works fine for plotting using motifStack instead of seqLogo.  I love the different plots in the motifStack Vignette.  

Our immediate need is to compare 2 pwms which DiffLogo is designed to do.  We can do this if all 20 AA's are present in the alignment as in the below example, but if the matrix contains less than 20 AA then it will not plot correctly.  Can stackMotif do something similar?

Thanks, Jeff

library(Biostrings)
library(seqLogo)
library(DiffLogo)

alignment1 <- c("ACDEEFGHIKLMNPQRSTVWY", "ACDEEFGHIKLMNPQRSTVWY", "ACDEEFGHIKLMNPQRSTVWY")
pwm1 <- consensusMatrix(alignment1, as.prob = TRUE)
seqLogo(pwm = pwm1, alphabet = ASN)

alignment2 <- c("ACDDEFGHIKLMNPQRSTVWY", "ACDDEFGHIKLMNPQRSTVWY", "ACDCEFGHIKLMNPQRSTVWY")
pwm2 <- consensusMatrix(alignment2, as.prob = TRUE)
seqLogo(pwm = pwm2, alphabet = ASN)

diffLogoFromPwm(pwm1 = pwm1, pwm2 = pwm2, alphabet = ASN)

Hey Jeff,

I tried to reproduce the bug, but I miss the function consensusMatrix. Could it be that the problem lies there? How does consensusMatrix know which alphabet to use?

I tried it with the DiffLogos internal method getPwmFromAlignment(alignment1, ASN, 0) and it worked fine.

Best regards

Martin