In IDG-Kinase/DarkKinaseTools: Tools for Interacting with Dark Kinase Data

#Load required libraries
library(tidyverse)
library(readxl)
library(here)
library(devtools)
#I know this is self referential, but I need to get access to some of the other
#data sets in this package to get the Hugo name mappings, I really want a more
#elegant way to do this
library(DarkKinaseTools)

Kinase.com Domains

Extract the kinase domains from the list maintained at kinase.com and determine the protein location within

kinome_com_file = here('data-raw','dark_kinases','kinase.com_list.xls')
if (! file.exists(kinome_com_file)) {
  download.file('http://kinase.com/human/kinome/tables/Kincat_Hsap.08.02.xls',
                kinome_com_file);
}

kinase_com_list = readxl::read_xls(kinome_com_file) %>%
  filter(`Pseudogene?` == "N") %>%
  select(Name,Protein,'Kinase Domain',KD2_Sequence) %>%
  rename(Kinase_domain = 'Kinase Domain') %>%
  #There appear to be extra characters in the kinase domain field, probably
  #related to other kinase domain alignments/problems with sequencing, let's
  #remove those
  mutate(Kinase_domain_clean = gsub('-','',Kinase_domain)) %>%
  mutate(Kinase_domain_clean = gsub('[*]','',Kinase_domain_clean)) %>%
  mutate(Protein_clean = gsub('[*]','',Protein)) %>%
  #Dealing with cases where there isn't a perfect match to a given kinase domain
  #on a protein by protein basis
  mutate(Kinase_domain_clean = case_when(
    Name == "HIPK4" ~ substr(Kinase_domain_clean,1,241),
    TRUE ~ Kinase_domain_clean))

kin_domain_locations = list();

for (kinase_row in 1:dim(kinase_com_list)[1]) {
  this_kin = kinase_com_list[kinase_row,]
  if (!is.na(this_kin$Kinase_domain)) {
    locate_hits = str_locate(this_kin$Protein_clean,this_kin$Kinase_domain_clean)
    #Make sure we found a hit for each domain
    if (length(locate_hits) == 0) {
      print(paste('No Hits found for:' ,this_kin$Name))
    }

    kin_domain_locations$kinase_com_name = c(kin_domain_locations$kinase_com_name,this_kin$Name)
    kin_domain_locations$start = c(kin_domain_locations$start,locate_hits[1])
    kin_domain_locations$stop = c(kin_domain_locations$stop,locate_hits[2])
    kin_domain_locations$protein_length = c(kin_domain_locations$protein_length,
                                            str_length(this_kin$Protein_clean))
  }

  if (!is.na(this_kin$KD2_Sequence)) {
    locate_hits = str_locate(this_kin$Protein_clean,this_kin$KD2_Sequence)

    if (length(locate_hits) == 0) {
      print(paste('No Hits found for:' ,this_kin$Name))
    }

    kin_domain_locations$kinase_com_name = c(kin_domain_locations$kinase_com_name,this_kin$Name)
    kin_domain_locations$start = c(kin_domain_locations$start,locate_hits[1])
    kin_domain_locations$stop = c(kin_domain_locations$stop,locate_hits[2])
    kin_domain_locations$protein_length = c(kin_domain_locations$protein_length,
                                            str_length(this_kin$Protein_clean))

  }
}

kin_domain_locations = as.data.frame(kin_domain_locations)

kin_domain_locations = left_join(kin_domain_locations,all_kinases)

devtools::use_data(kin_domain_locations, overwrite = TRUE)