scripts/parse_dto_results.R

library(tidyverse)
library(furrr)
library(here)

dto_results_lookup = expand_grid(
    pr_source = c("hackett", "hughes_oe", "hughes_ko", "hu_reimand", "kemmeren", "mahendrawada_rnaseq"),
    binding_source = c("cc", "chipexo", "harbison", "mahendrawada_chec"),
    pr_ranking_column = c("log2fc", "pvalue")) %>%
    mutate(
        result_file = map(file.path(here("results/dto"),
                                    pr_source,
                                    binding_source, "results",
                                    pr_ranking_column),
                          ~list.files(.x, full.names = TRUE))) %>%
    unnest(result_file) %>%
    mutate(tmp = str_remove(basename(result_file), ".json")) %>%
    separate_wider_delim(tmp,
                         names = c('binding_id', 'perturbation_id'),
                         delim = "-_-")

plan(multisession, workers = 20)

# Function with retry logic
read_json_with_retry = function(filepath, max_retries = 1) {
    for (attempt in 1:max_retries) {
        result = tryCatch({
            pivot_wider(enframe(as_vector(jsonlite::read_json(filepath))))
        }, error = function(e) {
            if (attempt < max_retries) {
                message(sprintf("Attempt %d/%d failed for %s: %s. Retrying...",
                                attempt, max_retries, filepath, e$message))
                Sys.sleep(2^(attempt - 1))  # Exponential backoff: 1s, 2s, 4s, 8s, 16s
                NULL
            } else {
                stop(sprintf("Failed after %d attempts for file: %s\nError: %s",
                             max_retries, filepath, e$message))
            }
        })

        if (!is.null(result)) {
            return(result)
        }
    }
}

dto_results_list = future_imap(dto_results_lookup$result_file, ~{
    tryCatch({
        # Check if file is empty
        if (file.size(.x) == 0) {
            return(tibble(
                empirical_pvalue = NA_real_,
                fdr = NA_real_,
                population_size = NA_real_,
                rank1 = NA_real_,
                rank2 = NA_real_,
                set1_len = NA_real_,
                set2_len = NA_real_,
                unpermuted_intersection_size = NA_real_,
                unpermuted_pvalue = NA_real_
            ))
        }
        read_json_with_retry(.x)
    }, error = function(e) {
        stop(sprintf("Index %d failed:\nFile: %s\nError: %s",
                     .y, .x, e$message))
    })
}, .progress = TRUE, .options = furrr_options(seed = TRUE))

dto_results_df = bind_cols(dto_results_lookup, bind_rows(dto_results_list))

hf_results_df = dto_results_df %>%
    mutate(
        binding_id = case_when(
            binding_source == "cc" & str_detect(binding_id, "-")
            ~ paste0("BrentLab/callingcards;annotated_features_combined;",
                     binding_id),
            binding_source == "cc" & str_detect(binding_id, "-", negate=TRUE)
            ~ paste0("BrentLab/callingcards;annotated_features;",
                     binding_id),
            binding_source == "chipexo"
            ~ paste0("BrentLab/rossi_2021;rossi_2021_af_combined;",
                     binding_id),
            binding_source == "harbison"
            ~ paste0("BrentLab/harbison_2004;harbison_2004;",
                     binding_id),
            binding_source == "mahendrawada_chec"
            ~ paste0("BrentLab/mahendrawada_2025;chec_mahendrawada_m2025_af_combined;",
                     binding_id)),
        perturbation_id = case_when(
            pr_source == "hackett" ~ paste0("BrentLab/hackett_2020;hackett_2020;", perturbation_id),
            pr_source == "hughes_oe" ~ paste0("BrentLab/hughes_2006;overexpression;", perturbation_id),
            pr_source == "hughes_ko" ~ paste0("BrentLab/hughes_2006;knockout;", perturbation_id),
            pr_source == "hu_reimand" ~ paste0("BrentLab/hu_2007_reimand_2010;hu_2007_reimand_2010;", perturbation_id),
            pr_source == "kemmeren" ~ paste0("BrentLab/kemmeren_2014;kemmeren_2014;", perturbation_id),
            pr_source == "mahendrawada_rnaseq" ~ paste0("BrentLab/mahendrawada_2025;rnaseq_reprocessed;", perturbation_id),
            .default = "ERROR"),
        binding_repo_dataset = case_when(
            binding_source == "cc" & str_detect(binding_id, "-")
            ~ "callingcards-annotated_features_combined",
            binding_source == "cc" & str_detect(binding_id, "-", negate=TRUE)
            ~ "callingcards-annotated_features",
            binding_source == "chipexo"
            ~ "rossi_2021-rossi_2021_af_combined",
            binding_source == "harbison"
            ~ paste0("harbison_2004-harbison_2004"),
            binding_source == "mahendrawada_chec"
            ~ "mahendrawada_2025-chec_mahendrawada_m2025_af_combined"),
        perturbation_repo_dataset = case_when(
            pr_source == "hackett" ~ paste0("hackett_2020-hackett_2020"),
            pr_source == "hughes_oe" ~ paste0("hughes_2006-overexpression"),
            pr_source == "hughes_ko" ~ paste0("hughes_2006-knockout"),
            pr_source == "hu_reimand" ~ paste0("hu_2007_reimand_2010-hu_2007_reimand_2010"),
            pr_source == "kemmeren" ~ paste0("kemmeren_2014-kemmeren_2014"),
            pr_source == "mahendrawada_rnaseq" ~ paste0("mahendrawada_2025-rnaseq_reprocessed"),
            .default = "ERROR")) %>%
    dplyr::rename(binding_rank_threshold = rank1,
                  perturbation_rank_threshold = rank2,
                  binding_set_size = set1_len,
                  perturbation_set_size = set2_len,
                  dto_fdr = fdr,
                  dto_empirical_pvalue = empirical_pvalue) %>%
    select(binding_id, perturbation_id,
           binding_rank_threshold, perturbation_rank_threshold,
           binding_set_size, perturbation_set_size,
           dto_fdr, dto_empirical_pvalue,
           pr_ranking_column,
           binding_repo_dataset, perturbation_repo_dataset)


# note that the deprecated directory is ignored in yeast_comparative_analysis.
# this exists on my computer and could exist on yours, too, if you wanted to
# put old results in a subdir called deprecated that would be ignored by huggingface/git
# old_db_dto_res = arrow::open_dataset("~/code/hf/yeast_comparative_analysis/deprecated/dto") %>%
#     collect()

# arrow::write_dataset(
#     hf_results_df,
#     path = "/home/chase/code/hf/yeast_comparative_analysis/dto",
#     format = "parquet",
#     partitioning = c("binding_repo_dataset", "perturbation_repo_dataset"),
#     existing_data_behavior = "overwrite",
#     compression = "zstd",
#     write_statistics = TRUE,
#     use_dictionary = c(
#         binding_id = TRUE,
#         perturbation_id = TRUE
#     )
# )