Package 'GExPipe'

Description

GExPipe users enter GSE accessions in the app; this helper runs automatically during GEO download when a count file has no sample headers (e.g. data.table::fread assigns V2, V3, ...). Columns are renamed to GSM IDs from GEO pData so QC, normalization, and group selection see individual samples (fixes issues such as a single V2 bar in outlier QC).

Usage

gexp_align_rnaseq_sample_names(count_matrix, metadata = NULL, gse_id = NULL)

Arguments

Value

Matrix with improved column names.

Examples

# GSE137136-style: headerless columns after download, GSM IDs in metadata.
mat <- matrix(1:4, nrow = 2, ncol = 2,
              dimnames = list(c("GENE1", "GENE2"), c("V2", "V3")))
meta <- data.frame(title = c("sample 1", "sample 2"),
                   row.names = c("GSM111", "GSM222"))
out <- gexp_align_rnaseq_sample_names(mat, meta, "GSE137136")
colnames(out)

Description

This helper wraps the Step 5 logic:

• filters genes by a variance percentile,

• applies one of several batch-correction methods,

• reports gene reduction statistics.

Usage

gexp_batch_correct(
  expr,
  metadata,
  variance_percentile = 25,
  method = c("combat_ref", "sva", "limma", "combat", "quantile_limma", "hybrid")
)

Arguments

Value

A list with elements:

Examples

expr <- matrix(rnorm(200), nrow = 20)
rownames(expr) <- paste0("Gene", seq_len(nrow(expr)))
colnames(expr) <- paste0("S", seq_len(ncol(expr)))
metadata <- data.frame(
  Dataset = rep(c("D1", "D2"), each = 5),
  Condition = rep(c("Normal", "Disease"), times = 5),
  row.names = colnames(expr),
  stringsAsFactors = FALSE
)
out <- gexp_batch_correct(expr, metadata, variance_percentile = 10, method = "limma")
dim(out$batch_corrected)

Description

Finalize common genes and combined matrix after download/mapping

Usage

gexp_download_finalize_common_genes(
  micro_expr_list,
  rna_counts_list,
  all_genes_list,
  subset_matrices = FALSE,
  keep_platforms_separate = FALSE
)

Arguments

Value

List with updated lists, common_genes, combined_expr_raw, and status.

Examples

m1 <- matrix(1:12, nrow = 3, dimnames = list(c("A", "B", "C"), paste0("S", 1:4)))
m2 <- matrix(1:12, nrow = 3, dimnames = list(c("B", "C", "D"), paste0("T", 1:4)))
out <- gexp_download_finalize_common_genes(
  micro_expr_list = list(D1 = m1),
  rna_counts_list = list(D2 = m2),
  all_genes_list = list(D1 = rownames(m1), D2 = rownames(m2))
)
out$common_genes

Description

Normalize dataset row IDs to gene symbols for overlap

Usage

gexp_download_normalize_ids_for_overlap(
  micro_expr_list,
  rna_counts_list,
  platform_per_gse = NULL,
  all_genes_list = NULL,
  micro_eset_list = NULL
)

Arguments

Value

List with updated lists and appended log text.

Examples

m <- matrix(1:6, nrow = 2, dimnames = list(c("A", "B"), paste0("S", 1:3)))
r <- matrix(1:6, nrow = 2, dimnames = list(c("B", "C"), paste0("T", 1:3)))
out <- gexp_download_normalize_ids_for_overlap(
  micro_expr_list = list(M = m),
  rna_counts_list = list(R = r)
)
names(out)

Description

Download and parse one microarray GSE

Usage

gexp_download_one_microarray_gse(gse_id, micro_dir, download_cel = NULL, fast = NULL)

Arguments

Value

List with status, log text, reason, expression, metadata, eset, platform_id, and cel_paths.

Examples

if (interactive()) {
  out <- gexp_download_one_microarray_gse("GSE10072", tempdir())
  out$ok
}

Description

Download and parse one RNA-seq GSE

Usage

gexp_download_one_rnaseq_gse(gse_id, rna_dir, fast = NULL)

Arguments

Value

List with status, reason, log text, count matrix, and metadata.

Examples

if (interactive()) {
  out <- gexp_download_one_rnaseq_gse("GSE50760", tempdir())
  out$ok
}

Description

Fetch sample metadata from GEO series matrix fallback

Usage

gexp_fetch_geo_series_matrix_metadata(gse_id)

Arguments

Value

data.frame or NULL.

Examples

if (interactive()) {
  md <- gexp_fetch_geo_series_matrix_metadata("GSE10072")
  if (!is.null(md)) head(md)
}

Description

Build diagnostic log text when no common genes are found

Usage

gexp_no_common_genes_diagnostic_log(all_genes_list)

Arguments

Value

Character string for log appending.

Examples

txt <- gexp_no_common_genes_diagnostic_log(
  list(D1 = c("1007_s_at", "1053_at"), D2 = c("ENSG000001", "ENSG000002"))
)
nchar(txt) > 0

Description

This is a non-Shiny helper version of the Step 2 (Normalize) logic:

• per-dataset normalization (microarray + RNA-seq),

• intersection of gene sets (common genes),

• global quantile normalization of the combined matrix,

• optional extraction of RNA-seq raw counts for count-based DE.

Usage

gexp_normalize_and_intersect(
  micro_expr_list,
  rna_counts_list,
  micro_norm_method = "quantile",
  rnaseq_norm_method = "TMM",
  micro_cel_paths = NULL,
  platform_per_gse = NULL,
  micro_eset_list = NULL,
  de_method = "limma",
  apply_global_quantile = TRUE,
  keep_platforms_separate = FALSE
)

Arguments

Value

A list with elements:

Examples

out <- withr::with_seed(1, {
  m1 <- matrix(abs(rnorm(120)), nrow = 20, ncol = 6)
  m2 <- matrix(abs(rnorm(120)), nrow = 20, ncol = 6)
  rownames(m1) <- rownames(m2) <- paste0("Gene", seq_len(20))
  colnames(m1) <- paste0("D1_S", seq_len(6))
  colnames(m2) <- paste0("D2_S", seq_len(6))
  gexp_normalize_and_intersect(
    micro_expr_list = list(D1 = m1, D2 = m2),
    rna_counts_list = list(),
    de_method = "limma"
  )
})
dim(out$combined_expr)

Description

RNA-seq and microarray boxes are always read. If both contain GSE IDs and the user did not choose Parallel, the run is treated as Merged. Parallel keeps both platforms separate through DE and accepts several GSEs per box.

Usage

gexp_parse_gse_inputs(
  analysis_type,
  rnaseq_gses = "",
  microarray_gses = "",
  dataset_mode = "multi"
)

Arguments

Value

List with rnaseq_ids, micro_ids, dataset_mode, analysis_type, and messages.

Examples

x <- gexp_parse_gse_inputs(
  analysis_type = "parallel",
  rnaseq_gses = "GSE1, GSE2",
  microarray_gses = "GSE3, GSE4",
  dataset_mode = "single"
)
x$rnaseq_ids

Description

Prepare clean download directories for current run

Usage

gexp_prepare_download_dirs(
  base_dir = getwd(),
  has_micro = FALSE,
  has_rna = FALSE,
  clear_cache = NULL
)

Arguments

Value

Character vector log lines describing cleanup.

Examples

td <- tempdir()
gexp_prepare_download_dirs(td, has_micro = TRUE, has_rna = TRUE)

Description

Used by QC outlier plots before normalization builds unified metadata.

Usage

gexp_qc_build_sample_dataset_map(micro_expr_list, rna_counts_list)

Arguments

Value

Named character vector: sample ID -> GSE accession.

Examples

m <- matrix(1:4, nrow = 2, ncol = 2, dimnames = list(c("A", "B"), c("S1", "S2")))
r <- matrix(1:4, nrow = 2, ncol = 2, dimnames = list(c("A", "B"), c("S3", "S4")))
gexp_qc_build_sample_dataset_map(list(GSE1 = m), list(GSE2 = r))

Description

Uses two methods on the top variable genes:

• PCA Mahalanobis distance (97.5% chi-square cutoff),

• Sample connectivity in signed network (power = 6; mean - 2*SD threshold).

Usage

gexp_qc_detect_outliers(expr, top_n = 5000L)

Arguments

Value

List with outlier detection results.

Examples

expr <- matrix(rnorm(400), nrow = 40, ncol = 10)
rownames(expr) <- paste0("Gene", seq_len(nrow(expr)))
colnames(expr) <- paste0("S", seq_len(ncol(expr)))
out <- gexp_qc_detect_outliers(expr, top_n = 20)
length(out$all_outliers)

Description

Combined-matrix PCA on mixed microarray + RNA-seq (or un-normalized multi-study counts) is not comparable. This helper runs gexp_qc_detect_outliers on each GSE separately and unions the flags.

Usage

gexp_qc_detect_outliers_per_dataset(
  micro_expr_list = list(),
  rna_counts_list = list(),
  top_n = 5000L
)

Arguments

Value

List with pca_outliers, conn_outliers, all_outliers, skipped (datasets with fewer than 5 samples), and log.

Examples

set.seed(1)
make_mat <- function(prefix) {
  m <- matrix(rnorm(400), nrow = 40, ncol = 10)
  dimnames(m) <- list(paste0("Gene", seq_len(40)), paste0(prefix, seq_len(10)))
  m
}
out <- gexp_qc_detect_outliers_per_dataset(
  micro_expr_list = list(GSE_A = make_mat("A")),
  rna_counts_list = list(GSE_B = make_mat("B")),
  top_n = 20
)
out$all_outliers

Description

Exclude selected samples from download/QC state lists

Usage

gexp_qc_exclude_samples(
  combined_expr_raw,
  micro_expr_list,
  rna_counts_list,
  unified_metadata = NULL,
  samples_to_exclude
)

Arguments

Value

List containing updated matrices/lists/metadata.

Examples

expr <- matrix(rnorm(300), nrow = 30, ncol = 10)
rownames(expr) <- paste0("Gene", seq_len(nrow(expr)))
colnames(expr) <- paste0("S", seq_len(ncol(expr)))
meta <- data.frame(SampleID = colnames(expr), stringsAsFactors = FALSE)
res <- gexp_qc_exclude_samples(
  combined_expr_raw = expr,
  micro_expr_list = list(D1 = expr[, 1:5, drop = FALSE]),
  rna_counts_list = list(D2 = expr[, 6:10, drop = FALSE]),
  unified_metadata = meta,
  samples_to_exclude = c("S1", "S10")
)
ncol(res$combined_expr_raw)

Description

Build gene-overlap summary table for QC UI

Usage

gexp_qc_gene_overlap_summary(all_genes_list, common_genes)

Arguments

Value

Data.frame with Dataset, Total_Genes, Pct_in_Common.

Examples

all_genes <- list(
  D1 = c("A", "B", "C", "D"),
  D2 = c("B", "C", "E")
)
gexp_qc_gene_overlap_summary(all_genes, common_genes = c("B", "C"))

Description

Prepare QC boxplot data from combined expression

Usage

gexp_qc_prepare_boxplot_data(
  combined_expr_raw,
  micro_expr_list,
  rna_counts_list,
  max_points = 500000L
)

Arguments

Value

data.frame with Expression, Sample, Platform.

Examples

expr <- matrix(rnorm(120), nrow = 12, ncol = 10)
rownames(expr) <- paste0("Gene", seq_len(nrow(expr)))
colnames(expr) <- paste0("S", seq_len(ncol(expr)))
df <- gexp_qc_prepare_boxplot_data(
  combined_expr_raw = expr,
  micro_expr_list = list(D1 = expr[, 1:5, drop = FALSE]),
  rna_counts_list = list(D2 = expr[, 6:10, drop = FALSE]),
  max_points = 1000
)
head(df)

Description

Prepare density curves for QC density plot

Usage

gexp_qc_prepare_density_data(combined_expr_raw, max_samples = 50L)

Arguments

Value

List with first density, additional densities, and colors.

Examples

expr <- matrix(rnorm(200), nrow = 20, ncol = 10)
d <- gexp_qc_prepare_density_data(expr, max_samples = 5)
length(d$others)

Description

Prepare UpSet matrix data from per-dataset genes

Usage

gexp_qc_prepare_upset_data(all_genes_list)

Arguments

Value

List with ok, upset_df, gene_lists, max_set_size, and message.

Examples

sets <- list(
  D1 = c("A", "B", "C"),
  D2 = c("B", "C", "D")
)
out <- gexp_qc_prepare_upset_data(sets)
names(out)

Description

Prepare cleaned gene sets for Venn plotting

Usage

gexp_qc_prepare_venn_sets(all_genes_list, max_sets = 5L)

Arguments

Value

List with ok, sets, and message.

Examples

sets <- list(
  D1 = c("A", "B", "C"),
  D2 = c("B", "C", "D"),
  D3 = c("A", "D")
)
out <- gexp_qc_prepare_venn_sets(sets)
out$ok

Description

Rebuild per-dataset gene lists from expression/count lists

Usage

gexp_rebuild_all_genes_list(micro_expr_list, rna_counts_list)

Arguments

Value

Named list of gene vectors.

Examples

m <- matrix(1:6, nrow = 2, dimnames = list(c("A", "B"), paste0("S", 1:3)))
r <- matrix(1:6, nrow = 2, dimnames = list(c("B", "C"), paste0("T", 1:3)))
gexp_rebuild_all_genes_list(list(M = m), list(R = r))

Description

Run differential expression analysis

Usage

gexp_run_de(
  expr,
  metadata,
  method = c("limma", "limma_voom", "deseq2", "edger"),
  logfc_cutoff = 0.5,
  padj_cutoff = 0.05
)

Arguments

Value

list with de_results, sig_genes, filter_note, sample_info, formula_desc

Examples

expr <- matrix(rnorm(200), nrow = 20)
rownames(expr) <- paste0("G", seq_len(nrow(expr)))
colnames(expr) <- paste0("S", seq_len(ncol(expr)))
meta <- data.frame(
  Condition = rep(c("Normal", "Disease"), each = 5),
  row.names = colnames(expr),
  stringsAsFactors = FALSE
)
gexp_run_de(expr, meta)

Description

This helper mirrors the "Prepare WGCNA Data" step:

• uses batch-corrected expression (or combined expression if needed),

• drops samples/genes with too many missing values,

• optionally filters to top variable genes,

• enforces a minimum fraction of non-missing samples per gene,

• returns a WGCNA-ready datExpr matrix (samples x genes) and companion sample info and gene-variance table.

Usage

gexp_wgcna_prepare(
  expr,
  metadata,
  gene_mode = c("top_variable", "all_common"),
  top_genes = 5000L,
  min_samples_frac = 0.5
)

Arguments

Value

A list with elements:

Examples

expr <- matrix(rnorm(600), nrow = 60)
rownames(expr) <- paste0("Gene", seq_len(nrow(expr)))
colnames(expr) <- paste0("S", seq_len(ncol(expr)))
metadata <- data.frame(
  SampleID = colnames(expr),
  Condition = rep(c("Normal", "Disease"), each = 5),
  Dataset = rep(c("D1", "D2"), each = 5),
  stringsAsFactors = FALSE
)
prep <- gexp_wgcna_prepare(expr, metadata, gene_mode = "top_variable", top_genes = 50)
dim(prep$datExpr)

Description

Build reproducibility report text for export

Usage

gexpipe_analysis_report_text(params = list(), include_session = TRUE)

Arguments

Value

Character vector of report lines.

Examples

gexpipe_analysis_report_text(list(method = "limma"), include_session = FALSE)

Description

Summarise Dataset x Condition confounding for batch/DE guidance

Usage

gexpipe_batch_confounding_summary(metadata)

Arguments

Value

list with confounded (logical), table (matrix), message (character)

Examples

meta <- data.frame(
  Dataset = rep(c("GSE1", "GSE2"), each = 3),
  Condition = rep(c("Normal", "Disease"), times = 3)
)
gexpipe_batch_confounding_summary(meta)

Description

Summarise how Platform should enter batch/DE models

Usage

gexpipe_batch_covariate_info(metadata)

Arguments

Value

list with mixed_platforms, platform_dataset_confounded, include_platform_covariate

Examples

meta <- data.frame(
  Dataset = rep(c("GSE1", "GSE2"), each = 2),
  Platform = rep(c("Microarray", "RNAseq"), each = 2)
)
gexpipe_batch_covariate_info(meta)

Description

Build ComBat / removeBatchEffect model matrix (biology to preserve)

Usage

gexpipe_build_batch_mod(metadata)

Arguments

Value

model matrix

Examples

meta <- data.frame(
  Condition = rep(c("Normal", "Disease"), each = 3),
  row.names = paste0("S", 1:6),
  stringsAsFactors = FALSE
)
gexpipe_build_batch_mod(meta)

Description

Build a model matrix for DE (limma / edgeR / voom)

Usage

gexpipe_build_de_design(metadata)

Arguments

Value

list with design, coef_condition, formula_desc, info

Examples

meta <- data.frame(
  Dataset = rep(c("GSE1", "GSE2"), each = 3),
  Condition = rep(c("Normal", "Disease"), times = 3),
  row.names = paste0("S", 1:6),
  stringsAsFactors = FALSE
)
gexpipe_build_de_design(meta)

Description

Keeps genes significant on both platforms. When require_same_direction is TRUE (default), the logFC sign must match.

Usage

gexpipe_consensus_degs(
  rna_sig,
  micro_sig,
  require_same_direction = TRUE,
  combine = c("intersection", "union")
)

Arguments

Details

Downstream tables use mean logFC and the more conservative (larger) adjusted p-value so WGCNA / common-gene steps can reuse sig_genes.

Value

list with table, genes, n_rna, n_micro, n_overlap, n_consensus, n_discordant, rna_only, micro_only.

Examples

rna <- data.frame(
  Gene = c("A", "B", "C"),
  logFC = c(1.2, -0.8, 1.1),
  adj.P.Val = c(0.01, 0.02, 0.03),
  Significance = c("Up-regulated", "Down-regulated", "Up-regulated"),
  stringsAsFactors = FALSE
)
micro <- data.frame(
  Gene = c("A", "B", "D"),
  logFC = c(0.9, -1.0, 1.4),
  adj.P.Val = c(0.02, 0.01, 0.04),
  Significance = c("Up-regulated", "Down-regulated", "Up-regulated"),
  stringsAsFactors = FALSE
)
out <- gexpipe_consensus_degs(rna, micro)
out$genes

Description

Summarise samples used in a DE run (transparency for mixed-platform runs)

Usage

gexpipe_de_sample_info(meta_used, total_meta = NULL, method = "limma")

Arguments

Value

list with human-readable fields for the Shiny UI.

Examples

meta <- data.frame(
  Condition = rep(c("Normal", "Disease"), each = 3),
  Platform = "RNAseq",
  row.names = paste0("S", 1:6),
  stringsAsFactors = FALSE
)
gexpipe_de_sample_info(meta)

Description

DESeq2 design formula for count-based DE

Usage

gexpipe_deseq2_design(metadata)

Arguments

Value

list with formula (formula object) and formula_desc (character)

Examples

meta <- data.frame(
  Condition = rep(c("Normal", "Disease"), each = 3),
  row.names = paste0("S", 1:6),
  stringsAsFactors = FALSE
)
gexpipe_deseq2_design(meta)

Description

Detect microarray + RNA-seq in the same analysis

Usage

gexpipe_has_mixed_platforms(metadata)

Arguments

Value

logical

Examples

meta <- data.frame(Platform = c("Microarray", "RNAseq"))
gexpipe_has_mixed_platforms(meta)

Description

Removes lowly expressed genes using a design-aware filter so filtering is not tied to differential expression statistics (avoids FDR bias from variance-percentile pre-filtering).

Usage

gexpipe_independent_filter(expr, design = NULL, group = NULL)

Arguments

Value

list with expr (filtered), keep (logical vector), n_before, n_after, note

Examples

expr <- matrix(stats::rpois(5000, lambda = 50), nrow = 500, ncol = 10)
design <- stats::model.matrix(~ factor(rep(c("A", "B"), each = 5)))
gexpipe_independent_filter(expr, design = design)

Description

Polar PCA coordinates for batch/platform diagnostic plots

Usage

gexpipe_pca_polar_df(expr, metadata, color_by = "Dataset")

Arguments

Value

data.frame with theta, r, and colour column.

Examples

expr <- matrix(rnorm(120), 20, 6, dimnames = list(paste0("G", 1:20), paste0("S", 1:6)))
meta <- data.frame(Dataset = rep(c("A", "B"), each = 3), row.names = colnames(expr))
gexpipe_pca_polar_df(expr, meta)

Description

Returns TRUE when Platform is nested within Dataset (each GSE has one technology) or when a tentative ~ Dataset + Platform (+ Condition) design is rank-deficient.

Usage

gexpipe_platform_dataset_confounded(metadata)

Arguments

Value

logical

Examples

meta <- data.frame(
  Dataset = c("GSE1", "GSE1", "GSE2", "GSE2"),
  Platform = c("Microarray", "Microarray", "RNAseq", "RNAseq")
)
gexpipe_platform_dataset_confounded(meta)

Description

Sample IDs for one platform in unified metadata

Usage

gexpipe_platform_sample_ids(metadata, platform = c("Microarray", "RNAseq"))

Arguments

Value

Character vector of sample IDs (may be empty).

Examples

meta <- data.frame(
  Platform = c("Microarray", "RNAseq"),
  row.names = c("S1", "S2")
)
gexpipe_platform_sample_ids(meta, "RNAseq")

Description

Aligns metadata to expression columns, runs PCA on samples, and estimates the mean fraction of variance in the top PCs explained by Dataset, Platform, and Condition. Residual is 1 - sum(factor fractions) (approximate).

Usage

gexpipe_pvca_df(expr, metadata, max_samples = 100L, max_pcs = 10L)

Arguments

Value

A list with ok (logical), data (data.frame or NULL), and message (character, empty on success).

Examples

expr <- matrix(rnorm(120), 20, 6, dimnames = list(paste0("G", 1:20), paste0("S", 1:6)))
meta <- data.frame(
  Dataset = rep(c("GSE1", "GSE2"), each = 3),
  Condition = rep(c("Normal", "Disease"), times = 3),
  row.names = colnames(expr),
  stringsAsFactors = FALSE
)
gexpipe_pvca_df(expr, meta)

Description

Uses per-platform normalized expression (not the cross-platform batch-corrected matrix). Single-dataset subsets use a Condition contrast; multiple datasets on the same platform use gexpipe_build_de_design().

Usage

gexpipe_run_limma_on_subset(
  expr,
  metadata,
  logfc_cutoff = 0.5,
  padj_cutoff = 0.05,
  ref_lab = "Normal",
  alt_lab = "Disease"
)

Arguments

Value

list with de_results, sig_genes, filter_note, sample_info, formula_desc.

Examples

expr <- matrix(rnorm(200), nrow = 20)
rownames(expr) <- paste0("G", seq_len(nrow(expr)))
colnames(expr) <- paste0("S", seq_len(ncol(expr)))
meta <- data.frame(
  Condition = rep(c("Normal", "Disease"), each = 5),
  row.names = colnames(expr),
  stringsAsFactors = FALSE
)
gexpipe_run_limma_on_subset(expr, meta)

Description

Detects your R version and verifies (or optionally installs) required packages. When GExPipe is installed from Bioconductor, dependencies are installed at package install time; gexpipe_setup() only checks availability unless options(gexpipe.auto_install = TRUE) is set.

Usage

gexpipe_setup(
  update = TRUE,
  optional = TRUE,
  launch = FALSE,
  port = 3838L,
  ...
)

Arguments

Value

Invisibly returns a named logical vector indicating which packages were successfully installed/available.

Examples

# Check and install missing packages only (no updates, no launch)
if (interactive()) {
  gexpipe_setup(update = FALSE, optional = FALSE, launch = FALSE)
}

Description

For a two-group design, WGCNA may add a combined "X vs Y" contrast column that duplicates one group indicator. This helper removes that column before plotting.

Usage

gexpipe_wgcna_heatmap_cor(cor_mat, combined)

Arguments

Value

The correlation matrix with the combined column removed when appropriate.

Examples

cor_mat <- matrix(runif(6), 2, 3, dimnames = list(c("ME1", "ME2"), c("A", "B", "A vs B")))
gexpipe_wgcna_heatmap_cor(cor_mat, "A vs B")

Description

Launches the GExPipe Shiny app for multi-omics RNA analysis (bulk RNA-seq, microarray, GEO download, QC, normalization, differential expression, WGCNA, pathway enrichment, PPI, machine learning, and more).

Usage

runGExPipe(
  launch.browser = TRUE,
  port = getOption("shiny.port", 3838),
  host = getOption("shiny.host", "127.0.0.1")
)

Arguments

Value

A Shiny app object. Users should run it with shiny::runApp(), which starts the server and blocks the R session until the app is stopped (e.g. RStudio Stop button).

Startup time

When GExPipe is installed from Bioconductor (or any normal R library), dependencies must be installed at package install time via BiocManager::install("GExPipe", dependencies = TRUE). runGExPipe() then verifies imports and opens the app without downloading packages.

For the GitHub / first-run workflow, set options(gexpipe.auto_install = TRUE) before runGExPipe() to enable background dependency installation (10-30 minutes on a fresh machine).

The welcome screen loads first; the full 15-tab dashboard is built only after the user clicks Go to Analysis, so the initial browser response is fast even on slow networks. Prefer GExPipe::runGExPipe() from an installed package rather than opening ⁠inst/shinyapp/⁠ in RStudio directly (that path runs global.R, which duplicates the install logic).

Optional tuning before runGExPipe():

• options(gexpipe.wgcna_threads = 1L) - fewer WGCNA threads (less parallel setup noise).

• options(gexpipe.wgcna_threads = 0L) or FALSE - skip WGCNA::enableWGCNAThreads() at app start.

• options(shiny.testmode = TRUE) with options(gexpipe.minimal_attach_in_testmode = TRUE) (default when unset) attaches only the Shiny stack first, then the rest after the first flush, so automated tests can connect quickly. Set gexpipe.minimal_attach_in_testmode = FALSE for a full attach on the first tick (e.g. deep shinytest2 scenarios).

PPI (STRINGdb)

Install with full dependencies so STRINGdb support packages resolve: BiocManager::install("GExPipe", dependencies = TRUE). STRINGdb is installed as a dependency. The first PPI run may download large STRING files (requires network). If initialization fails, try options(gexpipe.stringdb_try_versions = c("11.5", "11", "12")) or update STRINGdb.

Examples

# Non-interactive (R CMD check / example()): build an app object only.
# Dependency auto-install is skipped when interactive() is FALSE, so this
# runs quickly without touching the network.
app <- runGExPipe(launch.browser = FALSE, port = 0L)
stopifnot(inherits(app, "shiny.appobj"))

# Interactive: two-step launch. runGExPipe() builds the app object
# (installing any missing packages on first run); shiny::runApp() starts it.
if (interactive()) {
  app <- GExPipe::runGExPipe()      # Step 1: build the app object
  shiny::runApp(app)                # Step 2: start the server
}