design_new_okl.Rmd

---
title: "Design new OKL"
output: html_document
editor_options:
  chunk_output_type: console
---

```{r setup}
library(tidyverse)
library(synExtra)
library(here)
library(fst)
library(data.table)
library(powerjoin)
library(morgancpp)
# library(dtplyr)

theme_set(theme_minimal())

synapser::synLogin()
syn <- synDownloader("~/data/", .cache = TRUE)
```

```{r loading}
inputs <- list(
  okl_pseudo_kds = "syn51080578",
  fingerprints = "syn26262157"
)

input_files <- map(
  inputs, syn
)

pseudo_kds <- input_files[["okl_pseudo_kds"]] %>%
  fread() %>%
  as_tibble()

dir.create(here("new_okl"))
```


```{r}
fingerprints <- MorganFPS$new(input_files[["fingerprints"]], from_file = TRUE)
setdiff(pseudo_kds$lspci_id, fingerprints$names)

similarity_matrix <- fingerprints$tanimoto_subset(
  unique(single_dose_datasets$lspci_id),
  unique(single_dose_datasets$lspci_id)
) %>%
  as_tibble() %>%
  rename(lspci_id_1 = id_1, lspci_id_2 = id_2)

write_csv(
  similarity_matrix,
  here("new_okl", "similarity_values.csv")
)
```

Remove results from mutant kinases, then aggregate across remaining variants
(phosphorylation etc) using minimum pseudo-kd. Should hopefully reflect the
kinase state that people typically want to inhibit.


```{r}
kd_matrix <- pseudo_kds %>%
  filter(
    dataset == "original_repeat_replaced",
    !exclude_target
  ) %>%
  group_by(lspci_id, hgnc_symbol) %>%
  slice_min(pseudo_kd, na_rm = TRUE, with_ties = FALSE) %>%
  ungroup() %>%
  select(
    lspci_id, hmsl_id, name,
    `DiscoveRx Gene Symbol`, hgnc_symbol,
    classification, pseudo_kd, pseudo_kd_relation
  )

write_csv(
  kd_matrix,
  here("new_okl", "kd_values.csv")
)
```


```{r}
new_okl_syn <- synMkdir(
  "syn18508401", "new_okl"
)

synStoreMany(
  c(
    here("new_okl", "kd_values.csv"),
    here("new_okl", "similarity_values.csv")
  ),
  parentId = new_okl_syn,
  used = unname(inputs),
  executed = "https://github.com/labsyspharm/okl-analysis/blob/main/design_new_okl.Rmd",
  forceVersion = FALSE
)
```