Analyzing annotations in the Broad Drug Repurposing collection

Load sample information (updated 9/7/2018). Contains physical sample-level information including Broad Institute ID, compound name, QC confirmation, purity, vendor catalog number, vendor name, expected mass, SMILES, inChiKey, and Pubchem ID.

repurposing_samples <- 
  read_tsv("https://s3.amazonaws.com/data.clue.io/repurposing/downloads/repurposing_samples_20180907.txt", comment = "!", guess_max = 20000) %>%
  mutate(pert_id = str_sub(broad_id, 1, 13)) %>%
  select(-broad_id, -purity, -deprecated_broad_id, -vendor, -vendor_name, -catalog_no, -qc_incompatible) %>%
  distinct()

How many entries with duplicate pert_ids?

duplicate_pert_ids <- 
  repurposing_samples %>%
  group_by(pert_id) %>%
  tally() %>%
  arrange(desc(n)) %>%
  filter(n > 1) 

duplicate_pert_ids

nrow(duplicate_pert_ids)

[1] 16

Do the duplicates differ on any other parameter other than pert_iname? Zero rows means no.

duplicate_pert_ids %>%
  inner_join(repurposing_samples) %>%
  select(-pert_iname, -n) %>%
  distinct() %>%
  group_by(pert_id) %>%
  tally() %>%
  arrange(desc(n)) %>%
  filter(n > 1)

Load drug information (updated 9/7/2018). Contains drug annotation-level information including compound name, clinical phase, mechanism of action, and protein target.

repurposing_annotations <- 
  read_tsv("https://s3.amazonaws.com/data.clue.io/repurposing/downloads/repurposing_drugs_20180907.txt", comment = "!")

Make a table of all duplicates

duplicate_annotations <- 
  duplicate_pert_ids %>%
  inner_join(repurposing_samples) %>%
  select(pert_id, pert_iname) %>%
  distinct() %>%
  inner_join(repurposing_annotations)

Joining, by = "pert_id"
Joining, by = "pert_iname"

duplicate_annotations

Merge the annotations for each pert_id

merge_col <- function(column) {
  paste(column, collapse = "|") %>%
    str_split("\\|") %>%
    unlist %>%
    sort %>%
    unique %>%
    .[. != "NA"] %>%
    paste(collapse = "|")
}

duplicate_annotations %>%
  group_by(pert_id) %>%
  summarise_all(funs(merge_col))