IDR2D: Irreproducible Discovery Rate for Genomic Interactions

https://idr2d.mit.edu

Chromatin interaction data from protocols such as ChIA-PET and HiChIP provide valuable insights into genome organization and gene regulation, but can include spurious interactions that do not reflect underlying genome biology. We introduce a generalization of the Irreproducible Discovery Rate (IDR) method called IDR2D that identifies replicable interactions shared by experiments. IDR2D provides a principled set of interactions and eliminates artifacts from single experiments.

Installation

The idr2d package is part of Bioconductor since release 3.10. To install it on your system, enter:

if (!requireNamespace("BiocManager", quietly = TRUE)) {
  install.packages("BiocManager")
}

BiocManager::install("idr2d")

Alternatively, the development version can be installed directly from this repository:

if (!requireNamespace("remotes", quietly = TRUE)) {
  install.packages("remotes")
}

remotes::install_github("kkrismer/idr2d")

R 3.6 (or higher) and Bioconductor 3.10 (or higher) is required in both cases. Additionally, the 64-bit version of Python 3.5 (or higher) and the Python package hic-straw are required for Hi-C analysis from Juicer .hic files.

Usage

There are two vignettes available on Bioconductor, focusing on idr2d and ChIA-PET data and idr2d and ChIP-seq data.

The reference manual might also be helpful if you know what you are looking for.

Example code for ChiP-seq, ChIA-PET and Hi-C experiments

Analyzing results from replicate ChIP-seq experiments (stored in tab-delimited files chip-seq-rep1.txt and chip-seq-rep2.txt):

library(idr2d)

rep1_df <- read.table("chip-seq-rep1.txt", header = TRUE, sep = "\t",
                      stringsAsFactors = FALSE)
rep2_df <- read.table("chip-seq-rep2.txt", header = TRUE, sep = "\t",
                      stringsAsFactors = FALSE)

idr_results <- estimate_idr1d(rep1_df, rep2_df, 
                              value_transformation = "identity")
summary(idr_results)

rep1_idr_df <- idr_results$rep1_df
draw_idr_distribution_histogram(rep1_idr_df)
draw_rank_idr_scatterplot(rep1_idr_df)
draw_value_idr_scatterplot(rep1_idr_df)

Analyzing results from replicate ChIA-PET experiments (stored in tab-delimited files chia-pet-rep1.txt and chia-pet-rep2.txt):

library(idr2d)

rep1_df <- read.table("chia-pet-rep1.txt", header = TRUE, sep = "\t",
                      stringsAsFactors = FALSE)
rep2_df <- read.table("chia-pet-rep2.txt", header = TRUE, sep = "\t",
                      stringsAsFactors = FALSE)

idr_results <- estimate_idr2d(rep1_df, rep2_df, 
                              value_transformation = "identity")
summary(idr_results)

rep1_idr_df <- idr_results$rep1_df
draw_idr_distribution_histogram(rep1_idr_df)
draw_rank_idr_scatterplot(rep1_idr_df)
draw_value_idr_scatterplot(rep1_idr_df)

Analyzing chromosome 1 results in 1 Mbp resolution from replicate Hi-C experiments (stored in Juicer .hic files hic-rep1.hic and hic-rep2.hic):

library(idr2d)

rep1_df <- parse_juicer_matrix("hic-rep1.hic", resolution = 1e+06, chromosome = "chr1")
rep2_df <- parse_juicer_matrix("hic-rep2.hic", resolution = 1e+06, chromosome = "chr1")

idr_results_df <- estimate_idr2d_hic(rep1_df, rep2_df)
summary(idr_results_df)

draw_idr_distribution_histogram(idr_results_df)
draw_rank_idr_scatterplot(idr_results_df)
draw_value_idr_scatterplot(idr_results_df)
draw_hic_contact_map(idr_results_df, idr_cutoff = 0.05, chromosome = "chr1")

Analyzing chromosome 1 results in 1 Mbp resolution from replicate Hi-C experiments (stored in ICE normalized HiC-Pro .matrix and .bed files rep1_1000000_iced.matrix, rep1_1000000_abs.bed and rep2_1000000_iced.matrix, rep2_1000000_abs.bed):

library(idr2d)

rep1_df <- parse_hic_pro_matrix("rep1_1000000_iced.matrix", "rep1_1000000_abs.bed", chromosome = "chr1")
rep2_df <- parse_hic_pro_matrix("rep2_1000000_iced.matrix", "rep2_1000000_abs.bed", chromosome = "chr1")

idr_results_df <- estimate_idr2d_hic(rep1_df, rep2_df)
summary(idr_results_df)

draw_idr_distribution_histogram(idr_results_df)
draw_rank_idr_scatterplot(idr_results_df)
draw_value_idr_scatterplot(idr_results_df)
draw_hic_contact_map(idr_results, idr_cutoff = 0.05, chromosome = "chr1")

Build status

Platform	Status
Travis CI
Bioconductor 3.17 (release)
Bioconductor 3.18 (devel)

Citation

If you use IDR2D in your research, please cite:

IDR2D identifies reproducible genomic interactions
Konstantin Krismer, Yuchun Guo, and David K. Gifford
Nucleic Acids Research, Volume 48, Issue 6, 06 April 2020, Page e31; DOI: https://doi.org/10.1093/nar/gkaa030

Funding

The development of this method was supported by National Institutes of Health (NIH) grants 1R01HG008363 and 1R01NS078097, and the MIT Presidential Fellowship.