Skip to contents

Quick Start

This Quick Start runs proteoDA v2.0 on DIANN protein quant data without and with Perseus-style imputation, then writes per-contrast results, combined tables, and interactive plots.

Assumes your proteoDA_params.R defines: input_quan, metadata, contrasts, anno_start, anno_end, design, group, p.val, logFC, bin_size, filt_min_reps, require_both_groups, stat_cols, DA_dir, DA_table_cols, DA_title_col, ctrl_proteins, project_name, author, author2.

1. Load packages & parameters

library(proteoDA)
library(yaml)

# Common args from proteoDA_params.R
common_args <- list(
design_formula  = design,
pval_thresh     = p.val,
lfc_thresh      = logFC,
adj_method      = "BH",
binsize         = bin_size,
plot_movingSD   = TRUE,
contrasts_file  = contrasts
)

2. Read DIANN input and prepare IDS

stopifnot(file.exists(input_quan))
x <- read.csv(input_quan, stringsAsFactors = FALSE, check.names = FALSE)

if (!("Protein.Group" %in% colnames(x))) stop("Missing 'Protein.Group' in DIANN input.")

# Make uniprot_id from first entry of Protein.Group
u <- strsplit(x$Protein.Group, ";", fixed = TRUE)
uniprot_id <- vapply(u, function(z) if (length(z)) z[[1]] else NA_character_, "")
x <- cbind(uniprot_id = uniprot_id, x)

# Numeric NAs -> 0; character NAs/blanks -> "0"
for (nm in names(x)) {
  col <- x[[nm]]
if (is.numeric(col)) {
  col[is.na(col)] <- 0
} else {
  col[is.na(col) | col == ""] <- "0"
}
  x[[nm]] <- col
}

3. Metadata & annotation slice

stopifnot(file.exists(metadata))
md <- read.csv(metadata, stringsAsFactors = FALSE, check.names = FALSE)

if (!exists("anno_start") || !exists("anno_end")) stop("Define anno_start and anno_end in params.")
anno <- x[, anno_start:anno_end, drop = FALSE]

4. Align data to metadata

# Rename DIANN 'file' columns to 'sample' names where present
req <- c("file","sample")
mis <- req[!(req %in% colnames(md))]
if (length(mis)) stop(paste("Metadata missing:", paste(mis, collapse=", ")))

rename_map <- setNames(md$sample, md$file)
present <- intersect(names(rename_map), colnames(x))
if (length(present)) {
  idx <- match(present, colnames(x))
  colnames(x)[idx] <- rename_map[present]
}

# Keep sample columns in data that appear in metadata
keep <- colnames(x)[colnames(x) %in% md$sample]
intensity <- x[, keep, drop = FALSE]

# Trim metadata to those samples; rownames = sample
md <- md[md$sample %in% keep, , drop = FALSE]
rownames(md) <- md$sample

# Auto-align and validate
aln <- align_data_and_metadata(
    data                = intensity,
    metadata            = md,
    sample_col          = "sample",
    group_col           = "group",
    prefer_group_blocks = FALSE,
    strict              = FALSE
)
aln$changes  # optional preview of any reordering

intensity <- aln$data
md        <- aln$metadata

5. Build DAList and convert zeros → NA

raw <- DAList(
  data       = intensity,
  annotation = anno,
  metadata   = md,
  design     = NULL,
  eBayes_fit = NULL,
  results    = NULL,
  tags       = NULL
)

d0 <- zero_to_missing(raw)  # convert embedded zeros to NA for missing-aware ops
table(d0$metadata$group)
stopifnot(file.exists(contrasts))
flt <- filter_proteins_per_contrast(
    DAList              = d0,
    contrasts_file      = contrasts,
    min_reps            = filt_min_reps,
    require_both_groups = require_both_groups,
    grouping_column     = group
)

# Tag that we're using DIANN log2 normalized data
norm <- flt
norm$tags$norm_method <- "diann_quan"

# (Optional) quick counts to CSV
summary_df <- data.frame(
  contrast = names(flt$filtered_proteins_per_contrast),
  filtered_proteins = vapply(flt$filtered_proteins_per_contrast, length, 1L)
)
write.csv(summary_df, "filtered_protein_counts.csv", row.names = FALSE)

7. (Optional) Perseus-style imputation

set.seed(1)
imp <- perseus_impute(
    norm,
    shift = 1.8,
    width = 0.3,
    robust = TRUE,
    save_before_after = TRUE,
    seed = 1
)

# Save imputation diagnostics

invisible(write_perseus_imputation_plots(
    DAList     = imp,
    out_dir    = "PerseusPlots",
    bins       = 30,
    facet_ncol = 4,
    overlay    = TRUE,
    width      = 7,
    height     = 7,
    dpi        = 300,
    device     = "png"
))

8. Quick design sanity (full rank)

if (!exists("design")) stop("Define 'design' (e.g., ~0 + group) in params.")
mf <- model.frame(design, data = norm$metadata, drop.unused.levels = TRUE)
mm <- model.matrix(attr(mf, "terms"), data = mf)

cat("Columns:", ncol(mm), " Rank:", qr(mm)$rank, " Full rank?:", qr(mm)$rank == ncol(mm), "\n")

9. Run limma (no impute vs. Perseus)

common <- list(
  design_formula = design,
  pval_thresh    = p.val,
  lfc_thresh     = logFC,
  adj_method     = "BH",
  binsize        = bin_size,
  plot_movingSD  = TRUE,
  contrasts_file = contrasts
)

res_noimp <- do.call(run_filtered_limma_analysis, c(list(DAList = norm), common))
res_imp  <- do.call(run_filtered_limma_analysis, c(list(DAList = imp),  common))

10. Write tables & interactive plots

# Build statlists

sl_noimp <- build_statlist(DAList = res_noimp, stat_cols = stat_cols)
sl_imp   <- build_statlist(DAList = res_imp,   stat_cols = stat_cols)

# Tables (CSV + Excel with per-contrast sheets)
dir.create(DA_dir, showWarnings = FALSE, recursive = TRUE)
write_limma_tables(
    res_noimp,
    output_dir          = DA_dir,
    overwrite           = TRUE,
    contrasts_subdir    = NULL,
    summary_csv         = NULL,
    combined_file_csv   = NULL,
    spreadsheet_xlsx    = NULL,
    add_filter          = TRUE,
    color_palette       = NULL,
    add_contrast_sheets = TRUE, 
    statlist            = sl_noimp
)

dir.create("impute_out", showWarnings = FALSE, recursive = TRUE)
write_limma_tables(
    res_imp,
    output_dir             = "impute_out",
    overwrite              = TRUE,
    contrasts_subdir       = NULL,
    summary_csv            = NULL,
    combined_file_csv      = NULL,
    spreadsheet_xlsx       = NULL,
    add_filter             = TRUE,
    color_palette          = NULL,
    add_contrast_sheets    = TRUE,
    statlist               = sl_imp
)

# Interactive per-contrast HTML + static images

write_limma_plots(
    DAList           = res_noimp,
    grouping_column  = group,
    table_columns    = DA_table_cols,
    title_column     = DA_title_col,
    height           = 1000, width = 1000,
    output_dir       = DA_dir, overwrite = TRUE,
    control_proteins = ctrl_proteins,
    highlight_by     = "uniprot_id",
    image_formats    = c("pdf","png")
)

write_limma_plots(
    DAList           = res_imp,
    grouping_column  = group,
    table_columns    = DA_table_cols,
    title_column     = DA_title_col,
    height           = 1000, width = 1000,
    output_dir       = "impute_out", overwrite = TRUE,
    control_proteins = ctrl_proteins,
    highlight_by     = "uniprot_id",
    image_formats    = c("pdf","png")
)

11. Save DAList Objects for QC and Analysis Reports

# Without imputation (NAs)
saveRDS(res_noimp, "results_NA.rds")

# With Perseus imputation
saveRDS(res_imp, "results_NA_imputed.rds")

# Also provide a zero-filled version for certain QC/PCA utilities that require no NAs
results0 <- missing_to_zero(res_noimp)
saveRDS(results0, "results.rds")

results0_imputed <- missing_to_zero(res_imp)
saveRDS(results0_imputed, "results_imputed.rds")

12. Update _variables.yml for generation of PowerPoint summary

Update title/authors/project for downstream report generators

stopifnot(file.exists("_variables.yml"))
variables <- yaml.load_file("_variables.yml")

variables$title   <- project_name
variables$author  <- author
variables$author2 <- author2
variables$project <- project_name

write_yaml(variables, "_variables.yml")

# save R Session info
sink("sessionInfo.txt")
print(sessionInfo())
sink()
sessionInfo()
  • Results without imputation are under DA_dir/.
  • Results with Perseus imputation are under impute_out/.
  • Standalone per-contrast HTML reports are created in those folders.