Skip to contents

TangentXY

Overview

TangnetXY is a sex-informed copy number inference algorithm which extends Tangent (PMID: 36040167).

Installation

Dependencies

To install dependencies using R:

install.packages(c("readr", "dplyr", "tibble", "tidyr", "tidyselect", "pracma",
  "stringr", "ggplot2", "ggbeeswarm", "ggh4x", "gtools", "scales", "purrr", "rlang",
  "BiocManager"))
BiocManager::install(c("S4Vectors", "IRanges", "GenomicRanges", "DNAcopy"))

To install dependencies using conda or mamba:

conda create -n tangentxy -c r -c bioconda r-readr r-dplyr r-tibble r-tidyr r-tidyselect \
  r-pracma r-stringr r-ggplot2 r-ggbeeswarm r-ggh4x r-gtools r-scales r-purrr r-rlang \
  bioconductor-s4vectors bioconductor-iranges bioconductor-genomicranges bioconductor-dnacopy
conda activate tangentxy

TangnetXY

To install TangentXY, you’ll need either the remotes or devtools package:

install.packages("remotes")
remotes::install_github("beroukhim-lab/RtangentXY")

Usage

For more detailed function documentation, please see https://beroukhim-lab.github.io/RtangentXY/

The inputs to the tangent algorithm are as follows:

  1. Normal sample signal matrix with values in log2(Relative Copy Number) format
  2. Tumor sample signal matrix with values in log2(Relative Copy Number) format
  3. Probe information file
  4. Sample information file

The package comes with example datasets to demonstrate the expected format of each input.

Tangent 

library(RtangentXY)
library(dplyr)

# Filter out common germline CNVs
probes_filt <- filter_blacklist(example_pif)
nsig_df <- example_nsig_df %>% filter(locus %in% probes_filt$locus)
tsig_df <- example_tsig_df %>% filter(locus %in% probes_filt$locus)

# Diagnostic plot to select number of latent factors
plot_latent_factor_importance(
  sif_df  = example_sif,
  nsig_df = nsig_df,
  tsig_df = tsig_df
)

# Run tangent
n_latent <- 5
tangent_res <- run_tangent(
  sif_df   = example_sif,
  nsig_df  = nsig_df,
  tsig_df  = tsig_df,
  n_latent = n_latent
)

# Plot signal and noise profiles before and after normalization
plot_signal_noise(
  tnorm    = tangent_res,
  sif_df   = example_sif,
  pif_df   = probes_filt,
  tsig_df  = tsig_df,
  n_latent = n_latent
)

# Plot signal profile before and after tangent normalization for a given sample
plot_signal_along_loci(
  tnorm     = tangent_res,
  sample_id = "tumor.female1",
  pif_df    = probes_filt,
  tsig_df   = tsig_df,
  n_latent  = n_latent
)

Segmentation

The tangent pipeline outputs a normalized signal matrix which can then be used as input for Circular Binary Segmentation (CBS):

# Run CBS
cbs <- run_cbs(tangent_res, n_cores = 5)

# Save results to disk
save_segmentation(cbs, output_dir = ".")

Pseudotangent

When relatively few normal samples are available, the pseudotangent algorithm may be more appropriate.

ptangent_res <- run_pseudotangent(
  sif_df         = example_sif,
  nsig_df        = nsig_df,
  tsig_df        = tsig_df,
  n_latent_init  = 5,
  num_partitions = 3,
  n_latent_part  = 3
)

Support

Please open an issue for support.

Citation

These are the research papers associated with RtangentXY:

  • Gao GF, Oh C, Saksena G, Deng D, Westlake LC, Hill BA, Reich M, Schumacher SE, Berger AC, Carter SL, Cherniack AD, Meyerson M, Tabak B, Beroukhim R, Getz G. Tangent normalization for somatic copy-number inference in cancer genome analysis. Bioinformatics. 2022 Oct 14;38(20):4677-4686. doi: 10.1093/bioinformatics/btac586. PMID: 36040167; PMCID: PMC9563697.