Generating publication-ready sequence logos in R using ggseqlogo

Star

Watch

Issue

Download

Follow @omarwagih

Citation

If you use ggseqlogo please cite:

Wagih, Omar. ggseqlogo: a versatile R package for drawing sequence logos. Bioinformatics (2017). https://doi.org/10.1093/bioinformatics/btx469

Installation

First, install ggseqlogo from CRAN

install.packages("ggseqlogo")

or from github using the devtools package:

devtools::install_github("omarwagih/ggseqlogo")

Getting started

Load sample data

To get started, fire up the packages and load some sample data.

# Load the required packages
require(ggplot2)
require(ggseqlogo)

# Some sample data
data(ggseqlogo_sample)

This loads three sample data sets:

• seqs_dna: sets of binding sites for 12 transcription factors obtained from FASTA files in JASPAR. This is represented as a named list of character vectors, where the names represent the JASPAR ID.
• pfms_dna: a list of position frequency matrices for four transcription factors obtained from JASPAR. This is represented as a list of matrices, where the names represent the JASPAR ID.
• seqs_aa: sets of kinase-substrate phosphorylation sites obtained from Wagih et al. This is represented as a named list of character vectors where the names represent the names of the kinases associated with the phosphosites.

Plot a sequence logo

You can draw a sequence logos using ggplot function, with geom_logo. Let’s try this on sequences for one of the transcription factors from JASPAR:

ggplot() + geom_logo( seqs_dna$MA0001.1 ) + theme_logo()

You can also use the ggseqlogo as a shortcut to do the same thing. This is a wrapper function which adds theme_logo by default and performs any required faceting if multiple sequences logos are to be drawn. This is the function used throughout this tutorial.

ggseqlogo( seqs_dna$MA0001.1 )

Using the ggseqlogo wrapper function as a shortcut

Accepted input formats

ggseqlogo accepts three types of input, each described in detail below

1. Sequences: a character vector of aligned sequences
2. Matrices: a position frequency matrix, where the row is the letter, and column is the position. Note: the matrix must be row named with the letter. Matrices can also be a custom height matrix, but this is described later in the tutorial

The following generates a sequence logo using a position frequency matrix from the sample data

ggseqlogo( pfms_dna$MA0018.2 )

Creating sequence logos using position frequency matrices

Plotting methods

ggseqlogo supports two sequence logo methods through the method options: ‘bits’ and ‘probability’. By default, the bits is used.

p1 = ggseqlogo( seqs_dna$MA0001.1, method = 'bits' )
p2 = ggseqlogo( seqs_dna$MA0001.1, method = 'prob' )
gridExtra::grid.arrange(p1, p2)

Using different methods to plot a sequence logo

Sequence types

Preset alphabets

Amino acids, DNA and RNA sequence types are all supported by ggseqlogo. By default, ggseqlogo will try to guess your sequence type. You can explicitly set the sequence type through the seq_type option.

Lets try generate an amino acid sequence logo using kinase-substrate phosphorylation data:

ggseqlogo( seqs_aa$AKT1, seq_type='aa' )

An example sequence logo for amino acids

Custom alphabet

If you want to define a custom alphabet you can do so by setting namespace with your desired custom alphabet. For example, lets say you wanted a sequence logo of zeros and ones:

# Replace DNA characters with numbers
seqs_numeric = chartr('ATGC','1234', seqs_dna$MA0001.1)
ggseqlogo(seqs_numeric, method='p', namespace=1:4) 

Generating sequence logos with custom dictionaries - numerical characters

Greek letters are also supported:

# Replace DNA characters with Greek ones
seqs_greek = chartr('ATGC', 'δεψλ', seqs_dna$MA0001.1)
ggseqlogo(seqs_greek, namespace='δεψλ', method='p')

Generating sequence logos with custom dictionaries - Greek characters

For colors, you will need to create a custom color scheme for your namespace.

Colour schemes

Preset colour schemes

ggseqlogo has preset colour schemes that can be set using the col_scheme parameter. By default, the col_scheme is set to auto such that the colour scheme is automatically chosen based on your sequence type.

You can adjust the parameter. For amino acids you can pick from the following chemistry, hydrophobicity, clustalx, taylor. For DNA and RNA sequences nucleotide and base_pairing. For a full list of color schemes, see the list_col_schemes function. For example:

ggseqlogo(seqs_dna$MA0001.1, col_scheme='base_pairing')

An example of using a preset colour scheme

Custom colour schemes

If the presets are not enough for you, you can define custom discrete or continuous colour schemes using the make_col_scheme function. Here are two examples of discrete and continuous colour schemes.

Discrete color schemes

# Create custom colour scheme
cs1 = make_col_scheme(chars=c('A', 'T', 'C', 'G'), groups=c('gr1', 'gr1', 'gr2', 'gr2'), 
                      cols=c('purple', 'purple', 'blue', 'blue'))

# Generate sequence logo
ggseqlogo(seqs_dna$MA0001.1, col_scheme=cs1)

An example of a custom discrete colour scheme

Note that the groups parameter here is optional

Continuous color schemes

# Create custom colour scheme
cs2 = make_col_scheme(chars=c('A', 'T', 'C', 'G'), values=1:4)

# Generate sequence logo
ggseqlogo(seqs_dna$MA0001.1, col_scheme=cs2)

An example of a custom continuous colour scheme

Multiple sequence logos

You can plot more than one sequence logo at the same time with the help of facets. ggseqlogo will accept a named list of sequences or matrices. The names of the list will be used as the facet titles.

ggseqlogo(seqs_dna, ncol=4)

Generating multiple sequence logos at once using a list as an input

which is the same as calling:

ggplot() + geom_logo(seqs_dna) + theme_logo() + 
  facet_wrap(~seq_group, ncol=4, scales='free_x') 

Custom-height logos

If you have your own height metric for each letter, simply create a matrix where each cell is a the desired height, and set the method to custom. You can even have negative heights. Here’s a simple example:

# Create a custom matrix 
set.seed(123)
custom_mat = matrix( rnorm(20), nrow=4, dimnames=list(c('A', 'T', 'G', 'C')))

# Generate sequence logo
ggseqlogo(custom_mat, method='custom', seq_type='dna') + ylab('my custom height')

Using custom heights

Fonts

You can adjust the font by setting the font parameter. To list all the available color schemes use the list_fonts function.

fonts = list_fonts(F)

p_list = lapply(fonts, function(f){
  ggseqlogo(seqs_dna$MA0001.1, font=f) + ggtitle(f)
})

do.call(gridExtra::grid.arrange, c(p_list, ncol=2))

Using different fonts with ggseqlogo

Annotating logos

Overlaying annotation onto sequence logos is straightforward in ggseqlogo with ggplot2. Here is an example of drawing rectangles, lines and text.

ggplot() + 
  annotate('rect', xmin = 0.5, xmax = 3.5, ymin = -0.05, ymax = 1.9, alpha = .1, col='black', fill='yellow') +
  geom_logo(seqs_dna$MA0001.1, stack_width = 0.90) + 
  annotate('segment', x = 4, xend=8, y=1.2, yend=1.2, size=2) + 
  annotate('text', x=6, y=1.3, label='Text annotation') + 
  theme_logo()

Annotating sequence logos

Combining plots

Combining sequence logos with other plots generated by ggplot2 is simple. I’ll demonstrate with an example combining a sequence logo, sequence alignment and bar plot.

# Sequences we're going to use for the logo
seqs = seqs_dna$MA0008.1

# Generate the sequence logo
p1 = ggseqlogo(seqs) + theme(axis.text.x = element_blank())

# Make data for sequence alignment
aln = data.frame(
  letter=strsplit("AGATAAGATGATAAAAAGATAAGA", "")[[1]], 
  species = rep(c("a", "b", "c"), each=8),
  x       = rep(1:8, 3)
)
aln$mut = 'no'
aln$mut[ c(2,15,20,23) ] = 'yes'

# Generate the sequence alignment
p2 = ggplot(aln, aes(x, species)) +
  geom_text(aes(label=letter, color=mut, size=mut)) + 
  scale_x_continuous(breaks=1:10, expand = c(0.105, 0)) + xlab('') + 
  scale_color_manual(values=c('black', 'red')) + 
  scale_size_manual(values=c(5, 6)) + 
  theme_logo() + 
  theme(legend.position = 'none', axis.text.x = element_blank()) 

# Generate barplot data
bp_data = data.frame(x=1:8, conservation=sample(1:100, 8))

# Generate barplot data 
p3 = ggplot(bp_data, aes(x, conservation)) +
  geom_bar(stat='identity', fill='grey') + 
  theme_logo() + 
  scale_x_continuous(breaks=1:10, expand = c(0.105, 0)) + 
  xlab('')


# Now combine using cowplot, which ensures the plots are aligned
suppressMessages( require(cowplot) )
plot_grid(p1, p2, p3,  ncol = 1, align = 'v')

Combining sequence logos with other ggplot2 plots

Documentation

For more details on all features and parameters see ?ggseqlogo, ?geom_logo and ?make_col_scheme

Feedback

If you have any feedback or suggestions, please drop me a line at (omarwagih(at)gmail.com) or open an issue on Github.