happyR

Tools to help analyse your hap.py results in R. See the documentation for usage and examples.

Install

devtools::install_github("Illumina/happyR")

Demo

This example walks through a comparison of samples prepared using PCR-Free versus Nano library preps with 2 replicates per group.

library(happyR)
library(tidyverse, quietly = TRUE)

# groups are defined either by a CSV or data.frame with three 
# required columns: a label for each group (group_id), a unique
# label per replicate (replicate_id) and a path to the respective
# pre-computed hap.py output (happy_prefix)
extdata_dir <- system.file("extdata", package = "happyR")
# these extdata files are supplied with the package
samplesheet <- tibble::tribble(
  ~group_id,  ~replicate_id, ~happy_prefix,
  "PCR-Free", "NA12878-I30", paste(extdata_dir, "NA12878-I30_S1", sep = "/"),
  "PCR-Free", "NA12878-I33", paste(extdata_dir, "NA12878-I33_S1", sep = "/"),
  "Nano",     "NA12878-R1",  paste(extdata_dir, "NA12878-R1_S1", sep = "/"),
  "Nano",     "NA12878-R2",  paste(extdata_dir, "NA12878-R2_S1", sep = "/")
)

# here the above table is used to read hap.py output files from disk 
# and attach the group + replicate metadata
hap_samplesheet <- read_samplesheet_(samplesheet)

# extract summary PASS performance for each replicate and plot
summary <- extract_results(hap_samplesheet$results, table = "summary") %>% 
  inner_join(samplesheet, by = "happy_prefix") %>% 
  filter(Filter == "PASS")

ggplot(data = summary, aes(x = METRIC.Recall, y = METRIC.Precision, color = group_id, shape = Type)) +
  geom_point() + theme_minimal() + 
  xlim(NA, 1) + ylim(NA, 1) +
  scale_color_brewer(palette = "Set2") +
  labs(x = "Recall", y = "Precision", color = "Prep") +
  ggtitle("PCR-Free vs. Nano variant calling",
          "PCR treatment reduces indel calling performance")  

<img src="examples/README-usage-1.png" style="display: block; margin: auto;" />

System requirements

Originally developed for R v3.4.0. Tests are run using the most recent available R versions (incl. devel) on Ubuntu (Trusty) and OS X (El Capitan) platforms. happyR has not been tested on Windows. Dependencies are listed in DESCRIPTION.