scripts/cc_sra_submission_with_addtl.R

# All standard condition up to run_7390

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

# TODO: when making processing scripts github, add the background file
# and make sure the repo gets registered to zenodo. Put that link to the
# background files in the SRA submission

# cells grown on solid media at room temperature
# put the definition of the "Description" values in the

qc_django_data = read_csv("~/projects/parsing_yeast_database_data/data/qc_from_db/rr_20251222.csv")
expr_django_data = read_csv("~/projects/parsing_yeast_database_data/data/qc_from_db/expr_20251222.csv")
mcisaac_preferred_reps = expr_django_data %>%
    filter(source_name == "mcisaac_oe") %>%
    filter(preferred_replicate==TRUE) %>%
    pull(id)

qc_django_data_in_modeling = list(
    kemmeren = qc_django_data %>%
        filter(binding_source == "brent_nf_cc",
               !is.na(single_binding)) %>%
        filter(expression_source == "kemmeren_tfko"),
    mcisaac = qc_django_data %>%
        filter(binding_source == "brent_nf_cc",
               !is.na(single_binding)) %>%
        filter(expression_source == "mcisaac_oe"
                & expression %in% mcisaac_preferred_reps))

analysis_set_meta = read_csv("~/htcf_ref/data/yeast_database_modelling/pull_data_20250805/data/brent_nf_cc_meta_20250805.csv")
composite_binding = read_csv("~/htcf_ref/data/yeast_database_modelling/pull_data_20250805/data/bindingconcat_meta_20250805.csv") %>%
    filter(source_name == "brent_nf_cc")

barcode_details_root = "~/htcf_lts/sequence_data/yeast_cc/sequence"
barcode_details_list = list.files(barcode_details_root, "*_barcode_details.json",
                                  recursive=TRUE)

passing_bam_paths = read_csv("~/htcf_local/cc/yeast/passing_bams_lookup.txt", col_names = 'bampath')

passing_fastq_from_bam_paths = list.files("~/htcf_local/cc/yeast/passing_fastq_from_bam")

fastq_df = tibble(filename = passing_fastq_from_bam_paths) %>%
    extract(
        filename,
        into = c("batch", "regulator_symbol_replicate"),
        regex = "^(.+?)_([^_]+)_passing_tagged\\.fastq\\.gz$",
        remove = FALSE) %>%
    mutate(replicate = str_remove(str_extract(regulator_symbol_replicate, "x\\d"), "x")) %>%
    mutate(regulator_symbol = str_remove(regulator_symbol_replicate, "x\\d")) %>%
    replace_na(list(replicate = '1')) %>%
    mutate(replicate = as.integer(replicate)) %>%
    mutate(batch = ifelse(batch == "run_5690_correct", "run_5690", batch)) %>%
    filter(!batch %in% c('dsir4')) %>%
    filter(regulator_symbol != "undetermined") %>%
    # this is empty -- no passing hops
    filter(filename != "run_6739_MOT3_passing_tagged.fastq.gz")

barcode_details_df = map(barcode_details_list, ~{
    message(sprintf("working on %s", basename(.x)))
    x = jsonlite::read_json(file.path(barcode_details_root, .))
    tibble(
        seq = names(x$components$tf$map),
        regulator_symbol = unlist(x$components$tf$map)) %>%
        mutate(r1_index = substr(seq,1,5),
               r2_index = substr(seq,6,nchar(seq))) %>%
        mutate(batch = basename(dirname(.x)))}) %>%
    bind_rows()

composite_binding_unlisted = unlist(lapply(composite_binding$bindings, function(x) {
        as.numeric(str_extract_all(x, "\\d+")[[1]])}))

analysis_binding_ids = c(
    pull(filter(analysis_set_meta, !is.na(single_binding)), single_binding),
    composite_binding_unlisted
    )


binding_django = read_csv(here("data/binding_from_django_db_20260128.csv"))

barcode_details_df_with_id = barcode_details_df %>%
    mutate(replicate = str_remove(str_extract(regulator_symbol, "x\\d"), "x")) %>%
    mutate(condition = case_when(
        batch == "run_7380" & replicate == 2 ~ 'del_MET28',
        regulator_symbol == "CBF1KOmet28" ~ 'del_MET28',
        batch == "run_7380" & replicate == 2 ~ 'glu_1_gal_2',
        batch == "run_7388" & replicate == 2 ~ 'glu_1_gal_2',
        batch == "run_7390" & replicate == 2 ~ 'glu_1_gal_2',
        batch == "run_7392" & replicate == 2 ~ 'glu_1_gal_2',
        .default = 'standard')) %>%
    mutate(replicate = ifelse(condition != "standard", 1, replicate)) %>%
    mutate(regulator_symbol = ifelse(regulator_symbol == 'CBF1KOmet28', "CBF1", regulator_symbol)) %>%
    mutate(regulator_symbol = str_remove(regulator_symbol, "x\\d")) %>%
    replace_na(list(replicate = '1')) %>%
    mutate(replicate = as.integer(replicate)) %>%
    filter(batch != "run_5690") %>%
    mutate(batch = ifelse(batch == 'run_5690_correct', 'run_5690', batch)) %>%
    left_join(
        binding_django %>%
            select(id, regulator_symbol, batch, replicate)) %>%
    dplyr::rename(binding_id = id) %>%
    mutate(binding_id = as.character(binding_id)) %>%
    replace_na(list(binding_id = "NA"))

brentlab_dirs = list.dirs("~/htcf_lts/sequence_data/yeast_cc/sequence", recursive=FALSE)
mitra_dirs = list.dirs("~/htcf_lts/sequence_data/yeast_cc/sequence/mitra_data", recursive=FALSE)

setdiff(basename(c(brentlab_dirs, mitra_dirs)), unique(binding_django$batch))

gm_db = arrow::open_dataset("~/code/hf/callingcards/genome_map")

genome_map_meta_raw = arrow::read_parquet("~/code/hf/callingcards/genome_map_meta.parquet")

genome_map_meta = genome_map_meta_raw %>%
    filter(condition == "standard") %>%
    filter(!batch %in% c('dsir4'))

fastq_df_with_id = fastq_df %>%
    left_join(genome_map_meta %>%
                  mutate(regulator_symbol = ifelse(str_detect(regulator_symbol, "unknown"),
                                                   regulator_locus_tag,
                                                   regulator_symbol))) %>%
    filter(condition == 'standard') %>%
    filter(batch != "run_7392")

fastq_df_lookup = fastq_df_with_id %>%
    mutate(filename = file.path("passing_fastq_from_bam", filename),
           newname =  paste0(regulator_locus_tag, "_", regulator_symbol, "_", id)) %>%
    select(filename, newname)

bam_lookup = passing_bam_paths %>%
    mutate(base = str_remove(basename(bampath), ".bam")) %>%
    # this is empty, also excluded from the fastqs
    filter(base != "run_6739_MOT3_passing_tagged") %>%
    left_join(fastq_df_lookup %>%
                  mutate(base = str_remove(basename(filename), ".fastq.gz"))) %>%
    filter(str_detect(bampath, "undetermined", negate=TRUE)) %>%
    select(-filename) %>%
    # this removes the non standard conditions
    filter(complete.cases(.))

setdiff(bam_lookup$newname, fastq_df_lookup$newname)
setdiff(fastq_df_lookup$newname, bam_lookup$newname)

# bam_lookup %>%
#     select(bampath, newname) %>%
#     mutate(newname = paste0(newname, ".bam")) %>%
#     write_tsv("~/htcf_local/cc/yeast/passing_bam_rename_lookup.txt",
#               col_names = FALSE)

# fastq_df_lookup %>%
#     write_tsv("~/htcf_local/cc/yeast/passing_bam_fastq_rename_lookup.txt",
#               col_names = FALSE)


# fastq_df_with_id %>%
#     filter(binding_id == "NA") %>%
#     filter(condition == "standard") %>%
#     filter(regulator_symbol != "OTU1") %>%
#     mutate(qbed_path = file.path(sprintf("/home/chase/htcf_local/cc/yeast/results/%s/hops/%s_%s.qbed", batch, batch, regulator_symbol_replicate))) %>%
#     select(qbed_path) %>%
#     write_tsv("~/tmp/unprocessed_in_db_qbeds_lookup.txt")


# ACE2, ARG80, ARG81, LYS14, STB5, SWI5, SWI6 pass in at least one (didn't check which mcisaac cond pass, but all pass in kemmeren pass)
# CUP9 and DAL80 pass in mcisaac at 15 minutes
#
# WTM1, MIG2, DIG1 ACA1 fails due to non-passing in kemmeren and/or mcisaac. NOTE: the only timepoint MIG2 fails is 15 minutes.
# UME6 has no 15 minute condition in mcisaac and is not in kemmeren

af_django_meta = arrow::read_parquet("~/code/hf/callingcards/annotated_features_meta.parquet") %>%
    filter(condition == "standard") %>%
    filter(genome_map_id %in% genome_map_meta$id) %>%
    replace_na(list(binding_id = "NA")) %>%
    mutate(data_usable = case_when(
        batch %in% c('run_7477', 'run_7487')
        & regulator_symbol %in% c('ACE2', 'ARG80', 'ARG81', 'LYS14',
                                  'STB5', 'SWI5', 'SWI6', 'CUP9', 'DAL80') ~ "pass",
        batch %in% c('run_7477', 'run_7487')
        & regulator_symbol %in% c('WTM1', 'MIG2', 'DIG1', 'ACA1') ~ "fail",
        .default = data_usable)) %>%
    mutate(in_modeling_analysis = binding_id %in% analysis_binding_ids) %>%
    mutate(kemmeren_dto = binding_id %in%
               (qc_django_data_in_modeling$kemmeren %>%
               filter(dto_empirical_pvalue <= 0.01) %>%
               pull(single_binding))) %>%
    mutate(mcisaac_dto = binding_id %in%
               (qc_django_data_in_modeling$mcisaac %>%
               filter(dto_empirical_pvalue <= 0.01) %>%
               pull(single_binding))) %>%
    left_join(qc_django_data %>%
                  filter(!is.na(single_binding), binding_source == "brent_nf_cc") %>%
                  select(single_binding, genomic_inserts) %>%
                  distinct() %>%
                  mutate(single_binding = as.character(single_binding)) %>%
                  dplyr::rename(binding_id = single_binding)) %>%
    left_join(select(barcode_details_df_with_id, condition, regulator_symbol, batch, binding_id, r1_index, r2_index)) %>%
    dplyr::select(id, genome_map_id, batch,
                  r1_index, r2_index,
                  regulator_locus_tag, regulator_symbol,
                  data_usable, kemmeren_dto, mcisaac_dto, genomic_inserts,
                  in_modeling_analysis) %>%
    mutate(notes = case_when(
        genome_map_id %in% c(690, 685) ~ "manually exluded from analysis in favor of library 242",
        genome_map_id %in% c(26, 612, 300, 119) ~ "less than 3k insertions",
        .default = "none")) %>%
    arrange(genome_map_id)

af_db = arrow::open_dataset("~/code/hf/callingcards/annotated_features")

setdiff(basename(c(brentlab_dirs, mitra_dirs)), unique(genome_map_meta$batch))
setdiff(unique(genome_map_meta$batch),basename(c(brentlab_dirs, mitra_dirs)))


# db2506 = read_tsv("~/Downloads/2506.qbed.gz")
# hf2506 = gm_db %>%
#     filter(id == 1) %>%
#     collect()

# Specify output directory
# output_dir <- here("~/htcf_local/cc/yeast/callingcards_geo_submission/processed")
# dir.create(output_dir, showWarnings = FALSE, recursive = TRUE)

# Get unique combinations of id and batch
id_batch_combos <- af_django_meta %>%
    select(id, genome_map_id, batch) %>%
    distinct()

# for (i in 1:nrow(id_batch_combos)) {
#     current_id <- id_batch_combos$id[i]
#     current_batch <- id_batch_combos$batch[i]
#     gm_id = id_batch_combos$genome_map_id[i]
#
#     # Filter and format as af file
#     af_data <- af_db %>%
#         filter(id == current_id,
#                batch == current_batch) %>%
#         collect() %>%
#         mutate(genome_map_id = gm_id) %>%
#         select(-id) %>%
#         dplyr::rename(library_name = genome_map_id) %>%
#         mutate(target_symbol = ifelse(str_detect(target_symbol, "unknown"),
#                                       target_locus_tag, target_symbol)) %>%
#         dplyr::relocate(library_name, batch)
#
#     # Create filename
#     filename <- file.path(output_dir, paste0(gm_id, ".csv.gz"))
#
#     # Write gzipped csv file
#     write_csv(af_data, filename)
#
#     if (i %% 10 == 0) cat("Wrote", i, "of", nrow(id_batch_combos), "files\n")
# }
#
# cat("Done! Wrote", nrow(id_batch_combos), "CSV files to", output_dir, "\n")

af_data = arrow::open_dataset("~/code/hf/callingcards/annotated_features") %>%
    filter(id %in% af_django_meta$id) %>%
    collect() %>%
    left_join(
        af_django_meta %>%
            mutate(regulator_locus_tag = as.character(regulator_locus_tag),
                   regulator_symbol = as.character(regulator_symbol)) %>%
            mutate(regulator_symbol = ifelse(str_detect(regulator_symbol, "unknown"), regulator_locus_tag, regulator_symbol)) %>%
            mutate(filename = paste0(regulator_locus_tag, "_", regulator_symbol, "_", genome_map_id)) %>%
            select(id, genome_map_id, filename))

promoters = read_tsv("~/code/hf/yeast_genome_resources/yiming_promoters.bed",
                     col_names = c("chr", "start", "end", "target_locus_tag", "score", "strand")) %>%
    dplyr::select(target_locus_tag, chr, start, end, strand)

# af_data %>%
#     group_by(filename) %>%
#     group_walk(~ {
#         dir.create(here("results/processed"), showWarnings = FALSE)
#         output_name = paste0(.y$filename, ".csv.gz")
#         .x %>%
#             mutate(target_symbol = ifelse(str_detect(target_symbol, "unknown"), target_locus_tag, target_symbol)) %>%
#             dplyr::select(-c(id, hypergeometric_pval, batch)) %>%
#             dplyr::rename(enrichment = callingcards_enrichment) %>%
#             left_join(promoters) %>%
#             dplyr::relocate(genome_map_id, target_locus_tag, target_symbol, chr, start, end, strand) %>%
#         write_csv(file.path(here("results/processed"), output_name))
#     })


submission_df = af_django_meta %>%
    select(-id) %>%
    arrange(regulator_locus_tag) %>%
    mutate(regulator_locus_tag = as.character(regulator_locus_tag),
           regulator_symbol = as.character(regulator_symbol)) %>%
    mutate(regulator_symbol = ifelse(
        str_detect(regulator_symbol, "unknown"),
        regulator_locus_tag,
        regulator_symbol)) %>%
    mutate(`library name` = paste0(regulator_locus_tag, "_", regulator_symbol, "_", genome_map_id)) %>%
    mutate(title = paste0(regulator_locus_tag, " (", regulator_symbol, ") calling cards; gmid ", genome_map_id),
           `library strategy` = "OTHER",
           organism = 'Saccharomyces cerevisiae',
           `cell line` = "Saccharomyces cerevisiae S288C",
           molecule = 'genomic DNA',
           `single or paired-end` = 'single',
           `instrument model` = 'Illumina MiSeq',
           description = paste0(regulator_locus_tag, " tagged callingcards experiment. ",
                                "kemmeren_dto: ", kemmeren_dto, "; mcisaac_dto: ", mcisaac_dto,
                                "; in_modeling_analysis: ", in_modeling_analysis,
                                "; notes: ", notes),
           `processed data file` =  paste0(`library name`, ".csv.gz"),
           `raw file` = paste0(`library name`, ".bam")) %>%
    dplyr::rename(perturbation_validated = data_usable) %>%
    dplyr::select(
        `library name`, title, `library strategy`, organism, `cell line`,
        molecule, `single or paired-end`, `instrument model`,
        description, perturbation_validated,
        `processed data file`, `raw file`
    )

setdiff(paste0(bam_lookup$newname,".bam"), submission_df$`raw file`)
setdiff(submission_df$`raw file`, paste0(bam_lookup$newname,".bam"))

write_csv(submission_df, here("data/cc_submission_df.csv"))

################################################################################
################################################################################
################################################################################
# library(tidyverse)
# library(arrow)
# library(here)
#
# genome_map_meta = arrow::read_parquet("~/code/hf/callingcards/genome_map_meta.parquet")
#
# genome_map_meta_af = genome_map_meta %>% filter(batch %in% c("run_6778", "run_7380", "run_7388", "run_7390"))
#
# af = list.files("~/tmp/callingcards_output", full.names = TRUE)
#
# af_gmid_map = tibble(
#     af = str_remove(basename(af), "_af.csv"),
#     batch = str_extract(af, "run_\\d+"),
#     regulator_orig = str_remove(af, paste0(batch, "_")),
#     regulator_symbol = str_remove(regulator_orig, "x\\d")) %>%
#     mutate(regulator_symbol = ifelse(regulator_symbol == 'CBF1KOmet28', "CBF1", regulator_symbol)) %>%
#     mutate(condition = case_when(
#         str_detect(regulator_orig, "x\\d$", negate=TRUE) | str_detect(regulator_orig, "x1$") ~ 'standard',
#         regulator_orig == 'CBF1x2' ~ 'del_MET28',
#         regulator_orig == 'CBF1KOmet28' ~ 'del_MET28',
#         regulator_orig == 'GCR1x2' ~ 'glu_1_gal_2',
#         regulator_orig == 'GCR2x2' ~ 'glu_1_gal_2',
#         regulator_orig == 'TYE7x2' ~ 'glu_1_gal_2',
#         regulator_orig == "MIG1x2" ~ 'glu_1_gal_2')) %>%
#     left_join(dplyr::select(genome_map_meta_af, batch, regulator_symbol, condition, id)) %>%
#     mutate(id = ifelse(regulator_orig == 'CBF1KOmet28', 746, id))
#
#
# af_df = map(af, read_csv)
# names(af_df) = af_gmid_map$id
#
# genomicfeatures = arrow::read_parquet("~/code/hf/yeast_genome_resources/brentlab_features.parquet")
#
# promoters = read_tsv("~/code/hf/yeast_genome_resources/yiming_promoters.bed",
#                      col_names = c("chr", "start", "end", "name", "score", "strand")) %>%
#     dplyr::select("chr", "name", "start", "end", "strand") %>%
#     left_join(dplyr::select(genomicfeatures, name = locus_tag, symbol))
#
# af_df_all = bind_rows(af_df, .id = "genome_map_id") %>%
#     mutate(genome_map_id = as.integer(genome_map_id)) %>%
#     dplyr::select(c(
#         'genome_map_id','name','experiment_hops',
#         'background_hops','background_total_hops','experiment_total_hops',
#         'callingcards_enrichment','poisson_pval','hypergeometric_pval')) %>%
#     left_join(promoters) %>%
#     dplyr::rename(target_locus_tag = name, target_symbol = symbol) %>%
#     dplyr::select(
#         c('genome_map_id','target_locus_tag','target_symbol','experiment_hops',
#         'background_hops','background_total_hops','experiment_total_hops',
#         'callingcards_enrichment','poisson_pval','hypergeometric_pval')) %>%
#     left_join(dplyr::select(af_gmid_map, id, batch), by = c("genome_map_id" = "id")) %>%
#     mutate(genome_map_id = as.integer(genome_map_id))
#
# af_meta = arrow::read_parquet('~/code/hf/callingcards/annotated_features_meta.parquet')
#
# af_meta_new = af_gmid_map %>%
#     dplyr::select(id, batch, regulator_symbol, condition) %>%
#     dplyr::rename(genome_map_id = id) %>%
#     left_join(dplyr::select(genomicfeatures,
#                             regulator_locus_tag = locus_tag,
#                             regulator_symbol = symbol)) %>%
#     mutate(data_usable = 'unreviewed', analysis_set = FALSE) %>%
#     # note that this is 810 before adding these records
#     mutate(id = max(af_meta$id)+row_number(),
#            pss_id = "NA",
#            binding_id = "NA") %>%
#     dplyr::relocate(id)
#
# af_meta_augment = af_meta %>%
#     dplyr::select(-preferred_replicate) %>%
#     bind_rows(af_meta_new) %>%
#     replace_na(list(binding_id = "NA"))
#
# # af_meta_augment |>
# #     arrow::as_arrow_table() |>
# #     (\(tbl) {
# #         dict_cols <- c("data_usable", "batch", "condition", "regulator_locus_tag", "regulator_symbol")
# #         for (col in dict_cols) {
# #             tbl[[col]] <- tbl[[col]]$cast(arrow::dictionary())
# #         }
# #         tbl
# #     })() |>
# #     arrow::as_arrow_table() |>
# #     arrow::write_parquet(
# #         "/home/chase/code/hf/callingcards/annotated_features_meta.parquet",
# #         compression = "zstd",
# #         compression_level = 9,
# #         write_statistics = TRUE
# #     )
# #
# # af_df_all %>%
# #     left_join(dplyr::select(af_meta_new, id, genome_map_id)) %>%
# #     dplyr::relocate(id) %>%
# #     select(-genome_map_id) %>%
# #     arrow::write_dataset(
# #         path = "/home/chase/code/hf/callingcards/annotated_features_tmp",
# #         format = "parquet",
# #         partitioning = c("batch"),
# #         existing_data_behavior = "error",
# #         compression = "zstd",
# #         compression_level = 9,
# #         write_statistics = TRUE,
# #         use_dictionary = TRUE)