proposedAetiologyCOSMICv3.2: library(rvest)

In J0bbie/R2CPCT: Misc. functions to perform import and analysis of WGS and RNA-Seq samples

Description

# SBS Signatures ———————————————————-

Usage

data(proposedAetiologyCOSMICv3.2)

Format

Tibble containing the COSMIC (v3.2; March 2021) proposed signatures aetiologies.

Details

URL <- 'https://cancer.sanger.ac.uk/cosmic/signatures/SBS/'

# Retrieve content. x <- xml2::read_html(URL) html_nodes(css = '.vignettes_table_plot') html_children() html_text(trim = TRUE)

sigs.SBS <- rbind( do.call(rbind, lapply(x, function(s) s <- trimws(strsplit(s, '\n')[[1]]) data.frame(Signature = s[[1]], proposedAetiology = gsub('Proposed Aetiology', ”, s[[6]])) )) , data.frame(Signature = c('SBS27', 'SBS43', 'SBS45','SBS46','SBS47','SBS48','SBS49','SBS50','SBS51','SBS52','SBS53','SBS54','SBS55','SBS56','SBS57','SBS58','SBS59','SBS60'), proposedAetiology = 'Possible Sequencing Artifact') )

# DBS Signatures ———————————————————-

URL <- 'https://cancer.sanger.ac.uk/cosmic/signatures/DBS/'

# Retrieve content. x <- xml2::read_html(URL) html_nodes(css = '.vignettes_table_plot') html_children() html_text(trim = TRUE)

sigs.DBS <- do.call(rbind, lapply(x, function(s) s <- trimws(strsplit(s, '\n')[[1]]) data.frame(Signature = s[[1]], proposedAetiology = gsub('Proposed Aetiology', ”, s[[6]])) ))

# InDel Signatures ——————————————————–

URL <- 'https://cancer.sanger.ac.uk/cosmic/signatures/ID'

# Retrieve content. x <- xml2::read_html(URL) html_nodes(css = '.vignettes_table_plot') html_children() html_text(trim = TRUE)

sigs.ID <- do.call(rbind, lapply(x, function(s) s <- trimws(strsplit(s, '\n')[[1]]) data.frame(Signature = s[[1]], proposedAetiology = gsub('Proposed Aetiology', ”, s[[6]])) ))

# Combine lists ———————————————————–

proposedAetiologyCOSMICv3.2 <- rbind(sigs.SBS, sigs.DBS, sigs.ID)

# Group signatures based on common proposed aetiology together. proposedAetiologyCOSMICv3.2 <- proposedAetiologyCOSMICv3.2 dplyr::group_by(proposedAetiology) dplyr::mutate(proposedAetiologyGrouped = sprintf(' dplyr::ungroup()

usethis::use_data(proposedAetiologyCOSMICv3.2)

Source

COSMIC

J0bbie/R2CPCT documentation built on Feb. 24, 2022, 8:15 a.m.