#!/usr/bin/env Rscript

# Author: Denes Turei
# turei.denes@gmail.com
#
# Collect kinases and phosphatases from various resources
# using OmniPath and the OmnipathR R package
# https://github.com/saezlab/OmnipathR
#
# To see the list of resources: get_annotation_resources()

if(!('OmnipathR' %in% installed.packages())){

    require(devtools)
    devtools::install_github('saezlab/OmnipathR')

}

require(dplyr)
require(OmnipathR)

uniprot_kinases <-
    import_omnipath_annotations(
        resources = 'UniProt_keyword',
        wide = TRUE
    ) %>%
    filter(
        keyword == 'Kinase' &
        entity_type == 'protein'
    ) %>%
    pull(uniprot)


kinasecom_kinases <-
    import_omnipath_annotations(
        resources = 'kinase.com',
        entity_types = 'protein',
        wide = TRUE
    ) %>%
    pull(uniprot)


dgidb_kinases <-
    import_omnipath_annotations(
        resources = 'DGIdb',
        entity_types = 'protein',
        wide = TRUE
    ) %>%
    filter(
        category == 'KINASE'
    ) %>%
    pull(uniprot)


hgnc_kinases <- import_omnipath_annotations(
        resources = 'HGNC',
        wide = TRUE
    ) %>%
    filter(
        grepl('kinase', mainclass, ignore.case = TRUE) &
        entity_type == 'protein'
    ) %>%
    pull(uniprot) %>%
    unique()


uniprot_phosphatases <-
    import_omnipath_annotations(
        resources = 'UniProt_keyword',
        wide = TRUE
    ) %>%
    filter(
        keyword == 'Protein phosphatase' &
        entity_type == 'protein'
    ) %>%
    pull(uniprot)


hgnc_phosphatases <-
    import_omnipath_annotations(
        resources = 'HGNC',
        wide = TRUE
    ) %>%
    filter(
        grepl('rotein phosphatase', mainclass, fixed = TRUE) &
        entity_type == 'protein'
    ) %>%
    pull(uniprot) %>%
    unique()


dgidb_phosphatases <-
    import_omnipath_annotations(
        resources = 'DGIdb',
        entity_types = 'protein',
        wide = TRUE
    ) %>%
    filter(
        category == 'PROTEIN PHOSPHATASE'
    ) %>%
    pull(uniprot)


all_kinases <- c(
    uniprot_kinases,
    kinasecom_kinases,
    dgidb_kinases,
    hgnc_kinases
) %>% unique


all_phosphatases <- c(
    uniprot_phosphatases,
    hgnc_phosphatases,
    dgidb_phosphatases
) %>% unique