Why valr?

Why another tool set for interval manipulations? There are several other software packages available for genome interval analysis. However, based on our experiences teaching genome analysis, we were motivated to develop a toolset that:

  • Combines analysis and visualization in RStudio.
  • Can be used to generate reports with Rmarkdown.
  • Is highly extensible. New tools are quickly implemented on the R side.
  • Leverages the “modern R” syntax, using dplyr and the pipe operator from magrittr (%>%).
  • Maximizes speed by implementing compute-intensive algorithms in Rcpp.
  • Facilitates interactive visulaizations with shiny.

valr can currently be used for analysis of pre-processed data in BED and related formats. We plan to support BAM and VCF files soon via tabix indexes.

Input data

valr assigns common column names to facilitate comparisons between tbls. All tbls will have chrom, start, and end columns, and some tbls from multi-column formats will have additional pre-determined column names. See the read_bed() documentation for details.

valr can also operate on BED-like data.frames already constructed in R, provided that columns named chrom, start and end are present. New tbls can also be contructed using trbl_interval().

Interval coordinates

valr adheres to the BED format which specifies that the start position for an interval is zero based and the end position is one-based. The first position in a chromosome is 0. The end position for a chromosome is one position passed the last base, and is not included in the interval. For example:

Remote databases

Remote databases can be accessed with db_ucsc() (to access the UCSC Browser) and db_ensembl() (to access Ensembl databases).

Visual documentation

The bed_glyph() tool illustrates the results of operations in valr, similar to those found in the BEDtools documentation. This glyph shows the result of intersecting x and y intervals with bed_intersect():

And this glyph illustrates bed_merge():

Reproducible reports

valr can be used in RMarkdown documents to generate reproducible work-flows for data processing. Because valr is reasonably fast (see the benchmarks), we now use it in lieu of other tools for exploratory analysis of genomic data sets in R.

Command-line tools like BEDtools and bedops can be used in reproducible workflows (e.g., with snakemake), but it is cumbersome to move from command-line tools to exploratory analysis and plotting software. pybedtools can be used within ipython notebooks to accomplish a similar goal, but others have pointed out issues with this approach, including clunky version control. Because RMarkdown files are text files, they are readily kept under version control. Moreover, new features in RStudio (e.g. notebook viewing) enable similar functionality to ipython.

Grouping data

The group_by function in dplyr can be used to perform fuctions on subsets of single and multiple data_frames. Functions in valr leverage grouping to enable a variety of comparisons. For example, intervals can be grouped by strand to perform comparisons among intervals on the same strand.

Comparisons between intervals on opposite strands are done using the flip_strands() function:

Both single set (e.g. bed_merge()) and multi set operations will respect groupings in the input intervals.

Column specification

Columns in BEDtools are referred to by position:

In valr, columns are referred to by name and can be used in multiple name/value expressions for summaries.

# calculate the mean and variance for a `value` column
bed_map(a, b, .mean = mean(value), .var = var(value))

# report concatenated and max values for merged intervals
bed_merge(a, .concat = concat(value), .max = max(value))

Getting started

Meta-analysis

This demonstration illustrates how to use valr tools to perform a “meta-analysis” of signals relative to genomic features. Here we to analyze the distribution of histone marks surrounding transcription start sites.

First we load libraries and relevant data.

# `valr_example()` identifies the path of example files
bedfile <- valr_example('genes.hg19.chr22.bed.gz')
genomefile <- valr_example('hg19.chrom.sizes.gz')
bgfile  <- valr_example('hela.h3k4.chip.bg.gz')

genes <- read_bed(bedfile, n_fields = 6)
genome <- read_genome(genomefile)
y <- read_bedgraph(bgfile)

Then we generate 1 bp intervals to represent transcription start sites (TSSs). We focus on + strand genes, but - genes are easily accomodated by filtering them and using bed_makewindows() with reversed window numbers.

Now we use the .win_id group with bed_map() to caluclate a sum by mapping y signals onto the intervals in x. These data are regrouped by .win_id and a summary with mean and sd values is calculated.

Finally, these summary statistics are used to construct a plot that illustrates histone density surrounding TSSs.

library(ggplot2)

x_labels <- seq(-region_size, region_size, by = win_size * 5)
x_breaks <- seq(1, 41, by = 5)

sd_limits <- aes(ymax = win_mean + win_sd, ymin = win_mean - win_sd)

ggplot(res, aes(x = .win_id, y = win_mean)) +
  geom_point() + geom_pointrange(sd_limits) + 
  scale_x_continuous(labels = x_labels, breaks = x_breaks) + 
  xlab('Position (bp from TSS)') + ylab('Signal') + 
  ggtitle('Human H3K4me3 signal near transcription start sites') +
  theme_classic()

API

Function names are similar to their their BEDtools counterparts, with some additions.

Data types

Reading data

Transforming single interval sets

Comparing multiple interval sets

Randomizing intervals

Interval statistics

Utilities