scripts/parse_authors_data.R

library(tidyverse)
library(here)
library(arrow)
library(readxl)

# from fetchngs, the multiqc metadata has an easy way to map between
# accessions and samples
multqc_chec_config = yaml::read_yaml("~/Downloads/multiqc_config.yml")
chec_meta <- map_dfr(multqc_chec_config$sample_names_rename,
             ~tibble(!!!setNames(., multqc_chec_config$sample_names_rename_buttons)))

# genomic feature harmonization table ----
# see https://huggingface.co/datasets/BrentLab/yeast_genome_resources
genomicfeatures = arrow::open_dataset(here("data/genome_files/hf/features")) %>%
  as_tibble()

chec_data = read_excel(here("data/mahendrawada_2024_rnaseq/41586_2025_8916_MOESM5_ESM.xlsx"),
                         sheet="Table-S3b") %>%
  dplyr::rename(mahedrawada_target = `...1`) %>%
  pivot_longer(-mahedrawada_target,
               names_to = "mahedrawada_regulator_orig",
               values_to = "peak_score") %>%
  arrange(mahedrawada_regulator_orig, mahedrawada_target) %>%
  filter(!is.na(peak_score)) %>%
  mutate(mahedrawada_regulator = case_when(
    mahedrawada_regulator_orig == "MED15" ~ "GAL11",
    mahedrawada_regulator_orig == "YNR063W" ~ "PUL4",
    .default = mahedrawada_regulator_orig
  ))

rnaseq_data = read_excel(here("data/mahendrawada_2024_rnaseq/41586_2025_8916_MOESM5_ESM.xlsx"),
                         sheet="Table-S3c") %>%
  dplyr::rename(mahedrawada_target = `...1`) %>%
  pivot_longer(-c(mahedrawada_target, Kmeans_clusters),
               names_to = "mahedrawada_regulator_tmp",
               values_to = "log2fc") %>%
  arrange(mahedrawada_regulator_tmp, mahedrawada_target) %>%
  filter(!is.na(log2fc)) %>%
  separate(mahedrawada_regulator_tmp, into = c("mahedrawada_regulator_orig", "cond")) %>%
  replace_na(list(cond="SC")) %>%
  mutate(
    mahedrawada_regulator = case_when(
      mahedrawada_regulator_orig == "MED15" ~ "GAL11",
      mahedrawada_regulator_orig == "GALG4" ~ "GAL4",
      mahedrawada_regulator_orig == "YNR063W" ~ "PUL4",
      .default = mahedrawada_regulator_orig),
    cond = case_when(
      mahedrawada_regulator_orig == "GALG4" ~ "GAL",
      .default = cond
    ))

mahedrawada_genomicfeatures <- read_excel(
  here("data/mahendrawada_2024_rnaseq/41586_2025_8916_MOESM4_ESM.xlsx"),
  sheet = "ref_all"
) %>%
  mutate(gene_name = case_when(
    gene_name == "YPR022C" ~ "SDD4",
    gene_name == "YNR063W" ~ "PUL4",
    .default = gene_name
  ))

stopifnot(setequal(intersect(chec_data$mahedrawada_target, mahedrawada_genomicfeatures$gene_id),
                   chec_data$mahedrawada_target))

stopifnot(setequal(intersect(rnaseq_data$mahedrawada_target, mahedrawada_genomicfeatures$gene_id),
                   rnaseq_data$mahedrawada_target))

stopifnot(setequal(intersect(chec_data$mahedrawada_regulator, mahedrawada_genomicfeatures$gene_name),
                   chec_data$mahedrawada_regulator))

stopifnot(setequal(intersect(rnaseq_data$mahedrawada_regulator, mahedrawada_genomicfeatures$gene_name),
                   rnaseq_data$mahedrawada_regulator))

# note: verified by hand that where the symbol != gene_name, that the gene_name
# was listed as an alias of the current official symbol.
# Where the gene_name is == to the gene_id, that id is == to the locus_tag
stopifnot(
  tibble(mahedrawada_target = union(chec_data$mahedrawada_target,
                                    rnaseq_data$mahedrawada_target)) %>%
            left_join(genomicfeatures %>% select(locus_tag, symbol),
                      by = c("mahedrawada_target" = "locus_tag")) %>%
            left_join(mahedrawada_genomicfeatures %>% select(gene_name, gene_id),
                      by = c("mahedrawada_target" = "gene_id")) %>%
            filter(!complete.cases(.)) %>%
            nrow() == 0)

stopifnot(
  tibble(mahedrawada_regulator = union(chec_data$mahedrawada_regulator,
                                       rnaseq_data$mahedrawada_regulator)) %>%
  left_join(genomicfeatures %>%
              select(locus_tag, symbol),
            by = c("mahedrawada_regulator" = "symbol")) %>%
  left_join(mahedrawada_genomicfeatures %>%
              select(gene_name, gene_id),
            by = c("mahedrawada_regulator" = "gene_name")) %>%
    filter(!complete.cases(.) | locus_tag != gene_id) %>%
    nrow() == 0)

chec_data_final = chec_data %>%
  left_join(
    genomicfeatures %>%
      select(locus_tag, symbol) %>%
      mutate(mahedrawada_target = locus_tag) %>%
      dplyr::rename(target_locus_tag = locus_tag, target_symbol = symbol)) %>%
  left_join(
    genomicfeatures %>%
      select(locus_tag, symbol) %>%
      mutate(mahedrawada_regulator = symbol) %>%
      dplyr::rename(regulator_locus_tag = locus_tag, regulator_symbol = symbol)) %>%
  select(regulator_locus_tag, regulator_symbol, target_locus_tag, target_symbol, peak_score) %>%
  arrange(regulator_locus_tag, target_locus_tag)

rnaseq_data_final = rnaseq_data %>%
  left_join(
    genomicfeatures %>%
      select(locus_tag, symbol) %>%
      mutate(mahedrawada_target = locus_tag) %>%
      dplyr::rename(target_locus_tag = locus_tag, target_symbol = symbol)) %>%
  left_join(
    genomicfeatures %>%
      select(locus_tag, symbol) %>%
      mutate(mahedrawada_regulator = symbol) %>%
      dplyr::rename(regulator_locus_tag = locus_tag, regulator_symbol = symbol)) %>%
  select(regulator_locus_tag, regulator_symbol, target_locus_tag, target_symbol, log2fc) %>%
  arrange(regulator_locus_tag, target_locus_tag)

sample_id_map = tibble(
  regulator_locus_tag = union(rnaseq_data_final$regulator_locus_tag,
                                     chec_data_final$regulator_locus_tag)) %>%
  mutate(sample_id = row_number())


# chec_data_final %>%
#   left_join(sample_id_map) %>%
#   select(sample_id, all_of(colnames(chec_data_final))) %>%
#   write_parquet(here("~/code/hf/mahendrawada_2025/chec_mahendrawada_2025.parquet"),
#                 compression = "zstd",
#                 write_statistics = TRUE,
#                 use_dictionary = c(
#                   sample_id = TRUE,
#                   regulator_locus_tag = TRUE,
#                   regulator_symbol = TRUE,
#                   target_locus_tag = TRUE,
#                   target_symbol = TRUE
#                 )
#   )
#
# rnaseq_data_final %>%
#   left_join(sample_id_map) %>%
#   select(sample_id, all_of(colnames(rnaseq_data_final))) %>%
#   write_parquet(here("~/code/hf/mahendrawada_2025/rnaseq_mahendrawada_2025.parquet"),
#                 compression = "zstd",
#                 write_statistics = TRUE,
#                 use_dictionary = c(
#                   sample_id = TRUE,
#                   regulator_locus_tag = TRUE,
#                   regulator_symbol = TRUE,
#                   target_locus_tag = TRUE,
#                   target_symbol = TRUE
#                 )
#   )
#
# mahedrawada_genomicfeatures %>%
#   left_join(genomicfeatures %>%
#               select(locus_tag, symbol) %>%
#               mutate(gene_id = locus_tag)) %>%
#   write_parquet(here("~/code/hf/mahendrawada_2025/features_mahendrawada_2025.parquet"),
#                 compression = "zstd",
#                 write_statistics = TRUE,
#                 use_dictionary = c(
#                   gene_id = TRUE,
#                   gene_name = TRUE,
#                   chr = TRUE,
#                   locus_tag = TRUE,
#                   symbol = TRUE,
#                   coactivator = TRUE
#                 )
#   )