# library(plotly) #library(manhattanly) library(knitr) options(rmarkdown.html_vignette.check_title = FALSE) knitr::opts_chunk$set(echo = TRUE, eval=TRUE, tidy = FALSE, cache = FALSE, warning = FALSE, message = FALSE, dpi = 60) #comment = "#>") #knitr::opts_knit$set(eval.after = 'fig.cap')
Author: Sahir Bhatnagar (sahir.bhatnagar@gmail.com)
Notes:
This vignette was built with R markdown
and knitr
. The source code for this vignette can be found here.
In the interactive manhattan, qq and volcano plots below, we only use a subset (random sample of SNPs on chromosomes 4 to 7) of the HapMap
data included in this package. This is to reduce the size of the compiled vignette. To reduce the package size, we don't output any plots here. Please see https://sahirbhatnagar.com/manhattanly/articles/web_only/manhattanly_full.html for the full vignette with plots.
Manhattan, Q-Q and volcano plots are popular graphical methods for visualizing results from high-dimensional data analysis such as a (epi)genome wide asssociation study (GWAS or EWAS), in which p-values, Z-scores, test statistics are plotted on a scatter plot against their genomic position. Manhattan plots are used for visualizing potential regions of interest in the genome that are associated with a phenotype. Q-Q plots tell us about the distributional assumptions of the observed test statistics. Volcano plots are the negative log10 p-values plotted against their effect size, odds ratio or log fold-change. They are used to identify clinically meaningful markers in genomic experiments, i.e., markers that are statistically significant and have an effect size greater than some threshold.
Interactive manhattan, Q-Q and volcano plots allow the inspection of specific value (e.g. rs number or gene name) by hovering the mouse over a point, as well as zooming into a region of the genome (e.g. a chromosome) by dragging a rectangle around the relevant area.
This pacakge creates interactive Q-Q, manhattan and volcano plots that are usable from the R console, the RStudio
viewer pane, R Markdown
documents, in Dash
apps, Shiny
apps, embeddable in websites and can be exported as .png
files. By hovering the mouse over a point, you can see annotation information such as the SNP identifier and GENE name. You can also drag a rectangle to
zoom in on a region of interest and then export the image as a .png
file.
This work is based on the qqman
R package and the plotly.js
engine. It produces similar manhattan and Q-Q plots as the qqman::manhattan
and qqman::qq
functions; the main difference here is being able to interact with the plot, including extra annotation information and seamless integration with HTML.
You can install manhattanly
from CRAN:
install.packages("manhattanly")
Alternatively, you can install the development version of manhattanly
from GitHub with:
if (!require("pacman")) install.packages("pacman") pacman::p_load_gh("sahirbhatnagar/manhattanly")
The manhattanly
package ships with an example dataset called HapMap
. See help(HapMap)
for more details about how this dataset was created. Here is what the HapMap
dataset looks like:
# load the manhattanly library library(manhattanly) set.seed(12345) HapMap.subset <- subset(HapMap, CHR %in% 4:7) # for highlighting SNPs of interest significantSNP <- sample(HapMap.subset$SNP, 20) head(HapMap.subset) dim(HapMap.subset)
The required columns to create a manhattan plot are the chromosome, base-pair position and p-value. By default, the manhattanly
function assumes these columns are named CHR
, BP
and P
(but these can be specified by the user if they are different)
Create an interactive manhattan plot using one command:
manhattanly(HapMap.subset, snp = "SNP", gene = "GENE")
The arguments snp = "SNP"
and gene = "GENE"
specify that we want to add snp and gene information to each point. This information is found in the columns names "SNP"
and "GENE"
in the HapMap
dataset. See help(manhattanly)
for a full list of options.
Similarly, we can create an interactive Q-Q plot using one command (See help(qqly)
for a full list of options):
qqly(HapMap.subset, snp = "SNP", gene = "GENE")
You can then save the plot as a .png
file by clicking on the camera icon in the toolbar (which appears when you hover your mouse over it).
You can also make a volcano plot which by default, highlights the points greater than the default genomewideline
and effect_size_line
arguments:
volcanoly(HapMap.subset, snp = "SNP", gene = "GENE", effect_size = "EFFECTSIZE")
We can also highlight SNPs of interest using the highlight
argument. This package comes with a list of SNPs of interest called significantSNP
(see help(significantSNP)
for more details). To highlight these SNPs we simply pass this vector to the highlight
argument (note that these SNPs need to be present in the "SNP"
column of your data). We omit the plot here due to size constraints of the package on CRAN:
manhattanly(HapMap.subset, snp = "SNP", gene = "GENE", highlight = significantSNP)
You can add up to 4 annotations. In the following manhattan plot we add the snp, gene, the distance to the nearest gene and the effect size. The same annotations can also be added to Q-Q and volcano plots:
manhattanly(HapMap.subset, snp = "SNP", gene = "GENE", annotation1 = "DISTANCE", annotation2 = "EFFECTSIZE", highlight = significantSNP)
The annotations in the previous plots only appear when we hover the mouse over the point. Once we have identified a SNP, or a few SNPs of interest we want to explicitly show the annotation information and save the plot. The output of the manhattanly
function is an object which can be further manipulated using the %>%
operator from the magrittr
package:
library(magrittr) p <- manhattanly(HapMap.subset, snp = "SNP", gene = "GENE", annotation1 = "DISTANCE", annotation2 = "EFFECTSIZE", highlight = significantSNP) # get the x and y coordinates from the pre-processed data plotData <- manhattanr(HapMap.subset, snp = "SNP", gene = "GENE")[["data"]] # annotate the smallest p-value annotate <- plotData[which.min(plotData$P),] # x and y coordinates of SNP with smallest p-value xc <- annotate$pos yc <- annotate$logp p %>% plotly::layout(annotations = list( list(x = xc, y = yc, text = paste0(annotate$SNP,"<br>","GENE: ",annotate$GENE), font = list(family = "serif", size = 10))))
You can then save the plot as a .png
file by clicking on the camera icon in the toolbar (which appears when you hover your mouse over it).
By default, plotly for R runs locally in your web browser or in the R Studio viewer. It would be useful to be able to easily share these plot with for example, your collaborators or supervisor, especially in the during the exploratory data analysis stage of your project.
You can publish your charts to the web with plotly's web service in three steps.
Create a free plotly account here. A plotly account is required to publish charts online. It's free to get started, and you control the privacy of your charts.
Find your authentication API keys in your online settings. Set them in your R session with:
Sys.setenv("plotly_username"="your_plotly_username") Sys.setenv("plotly_api_key"="your_api_key")
plotly::api_create()
library(plotly) # p is the interactive manhattan plot we saved earlier plotly::api_create(p, filename = "r-docs/manhattan", world_readable=TRUE)
filename
sets the name of the file inside your online plotly account.world_readable
sets the privacy of your chart. If TRUE
, the graph is publically viewable, if FALSE
, only you can view it.knitr
and R Markdown
R Markdown
is a an R
software package that allows the creation of dynamic documents, i.e., embed R
code with text to create fully reproducible reports. Furthermore it allows easy creation of HTML
reports without knowing how to code in HTML
(such as this vignette). This means you can embed interactive manhattan, qq and volcano plots in HTML
reports using the manhattanly
package. For example, to embed the above manhattan plot I included the following code chunk in the .Rmd
document:
```r
library(plotly)
manhattanly(subset(HapMap, CHR %in% 4:7), snp = "SNP", gene = "GENE")
```
The manhattanly
package splits up the data pre-processing from the rendering of the plot object (inspired by the heatmaply
package. These are done by the manhattanr
, qqr
and volcanor
functions:
# create an object of class `manhattanr` manhattanrObject <- manhattanr(HapMap) # whats in there str(manhattanrObject) # the data used for plotting is a data.frame # this data.frame can be used with any graphics function such as in base R # you do not need to use plotly head(manhattanrObject[["data"]]) is.data.frame(manhattanrObject[["data"]])
This manhattanrObject
which is of class manhattanr
can also be passed to the manhattanly
function:
manhattanly(manhattanrObject)
We can specify more annotations in the data using the snp
, gene
, annotation1
and annotation2
arguments:
# create an object of class `manhattanr` manhattanrObject <- manhattanr(HapMap, snp = "SNP", gene = "GENE", annotation1 = "DISTANCE", annotation2 = "EFFECTSIZE") # the annotation columns are now part of the data.frame head(manhattanrObject[["data"]]) is.data.frame(manhattanrObject[["data"]])
Similarly the data used for the Q-Q plot can be created using the qqr
function:
qqrObject <- qqr(HapMap) str(qqrObject) head(qqrObject[["data"]])
This qqrObject
which is of class qqr
can also be passed to the qqly
function (we omit the plot here for the sake of size of the rendered vignette):
qqly(qqrObject)
Similarly the data used for the volcanor plot can be created using the volcanor
function:
volcanorObject <- volcanor(HapMap, p = "P", effect_size = "EFFECTSIZE", snp = "SNP", gene = "GENE") str(volcanorObject) head(volcanorObject[["data"]])
This can be directly passed to the volcanoly
function, with additional arguments:
volcanoly(volcanorObject, effect_size_line = c(-1, 0.5))
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.