Package 'GExPipe'

Align RNA-seq count-matrix column names with GEO sample metadata

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

count_matrix	Numeric matrix (genes x samples).
metadata	Optional GEO pData with sample IDs as row names.
gse_id	GEO series accession (used for fallback naming).

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)

---

Variance-based gene filtering and batch correction

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

expr	Combined expression matrix (genes x samples), after normalization.
metadata	Data.frame with at least columns Dataset and Condition.
variance_percentile	Numeric between 0 and 50; bottom percentile to remove.
method	Batch method: one of "limma", "combat", "combat_ref", "quantile_limma", "hybrid", "sva".

Value

A list with elements:

expr_filtered	filtered expression matrix before batch correction
batch_corrected	batch-corrected matrix
genes_before	integer, genes before filtering
genes_after	integer, genes after filtering
filter_percent	numeric, percent genes removed
log_text	character, human-readable log string

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)

---

Finalize common genes and combined matrix after download/mapping

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
)

Arguments

micro_expr_list	Named list of microarray matrices (genes x samples).
rna_counts_list	Named list of RNA-seq matrices (genes x samples).
all_genes_list	Named list of rowname vectors per dataset.

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

---

Normalize dataset row IDs to gene symbols for overlap

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

micro_expr_list	Named list of microarray matrices.
rna_counts_list	Named list of RNA-seq matrices.
platform_per_gse	Optional named list/vector of GPL IDs.
all_genes_list	Optional prebuilt gene lists.
micro_eset_list	Optional named list of ExpressionSet objects for fData fallback when probe-to-symbol mapping needs platform annotation.

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)

---

Download and parse one microarray GSE

Description

Download and parse one microarray GSE

Usage

gexp_download_one_microarray_gse(gse_id, micro_dir)

Arguments

gse_id	GEO series ID.
micro_dir	Directory for supplementary files.

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
}

---

Download and parse one RNA-seq GSE

Description

Download and parse one RNA-seq GSE

Usage

gexp_download_one_rnaseq_gse(gse_id, rna_dir)

Arguments

gse_id	GEO series ID.
rna_dir	Directory containing rna_data.

Value

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

Examples

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

---

Fetch sample metadata from GEO series matrix fallback

Description

Fetch sample metadata from GEO series matrix fallback

Usage

gexp_fetch_geo_series_matrix_metadata(gse_id)

Arguments

gse_id

GEO series ID.

Value

data.frame or NULL.

Examples

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

---

Build diagnostic log text when no common genes are found

Description

Build diagnostic log text when no common genes are found

Usage

gexp_no_common_genes_diagnostic_log(all_genes_list)

Arguments

all_genes_list

Named list of row IDs per dataset.

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

---

Normalize microarray and RNA-seq datasets and compute common genes

Description

This is a non-Shiny helper version of the Step 3 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
)

Arguments

micro_expr_list	named list of microarray expression matrices.
rna_counts_list	named list of RNA-seq count matrices.
micro_norm_method	"quantile" or "rma" (RMA requires CEL paths).
rnaseq_norm_method	"TMM" or "log2cpm_only".
micro_cel_paths	optional named list of CEL paths per GSE (for RMA).
platform_per_gse	optional named vector giving platform IDs per GSE.
micro_eset_list	optional named list of ExpressionSet objects per GSE (for RMA ID mapping).
de_method	differential expression method (used to decide whether to save raw counts for DESeq2/edgeR/limma-voom); one of "limma", "limma_voom", "deseq2", "edger".
apply_global_quantile	logical; if TRUE, apply limma quantile normalization across all samples.

Value

A list with elements:

combined_expr	globally quantile-normalized matrix (genes x samples)
combined_expr_before_global	matrix before global quantile
all_expr_norm_list	normalized per-dataset matrices (common genes only)
common_genes	character vector of common genes
normalization_stats	list of per-dataset normalization info
normalization_summary_table	data.frame summarizing gene counts
raw_counts_for_deseq2	integer matrix of RNA-seq counts (optional)
raw_counts_metadata	data.frame with sample metadata for raw counts (optional)
unified_metadata	data.frame with SampleID, Platform, Dataset, Condition=NA
log_text	character string with a human-readable log

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)

---

Parse GSE IDs from Shiny text inputs

Description

Parse GSE IDs from Shiny text inputs

Usage

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

Arguments

analysis_type	One of "rnaseq", "microarray", or "merged".
rnaseq_gses	Text field for RNA-seq IDs.
microarray_gses	Text field for microarray IDs.
dataset_mode	"single" or "multi".

Value

List with rnaseq_ids, micro_ids, dataset_mode, and messages.

Examples

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

---

Prepare clean download directories for current run

Description

Prepare clean download directories for current run

Usage

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

Arguments

base_dir	Base working directory.
has_micro	Logical; whether microarray IDs exist.
has_rna	Logical; whether RNA-seq IDs exist.

Value

Character vector log lines describing cleanup.

Examples

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

---

Map combined-expression sample IDs to source GSE datasets

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

micro_expr_list	Named list of microarray matrices.
rna_counts_list	Named list of RNA-seq matrices.

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))

---

Detect sample outliers from expression matrix

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

expr	Numeric matrix with genes in rows and samples in columns.
top_n	Integer number of top variable genes to use (default 5000).

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)

---

Exclude selected samples from download/QC state lists

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

combined_expr_raw	Matrix genes x samples.
micro_expr_list	Named list of microarray matrices.
rna_counts_list	Named list of RNA-seq matrices.
unified_metadata	Optional metadata data.frame with SampleID column.
samples_to_exclude	Character vector of sample IDs.

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)

---

Build gene-overlap summary table for QC UI

Description

Build gene-overlap summary table for QC UI

Usage

gexp_qc_gene_overlap_summary(all_genes_list, common_genes)

Arguments

all_genes_list	Named list of gene vectors by dataset.
common_genes	Character vector of common genes.

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"))

---

Prepare QC boxplot data from combined expression

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

combined_expr_raw	Matrix genes x samples.
micro_expr_list	Named list of microarray matrices.
rna_counts_list	Named list of RNA-seq matrices.
max_points	Maximum rows to keep for plotting.

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)

---

Prepare density curves for QC density plot

Description

Prepare density curves for QC density plot

Usage

gexp_qc_prepare_density_data(combined_expr_raw, max_samples = 50L)

Arguments

combined_expr_raw	Matrix genes x samples.
max_samples	Maximum number of sample density lines to include.

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)

---

Prepare UpSet matrix data from per-dataset genes

Description

Prepare UpSet matrix data from per-dataset genes

Usage

gexp_qc_prepare_upset_data(all_genes_list)

Arguments

all_genes_list

Named list of gene vectors by dataset.

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)

---

Prepare cleaned gene sets for Venn plotting

Description

Prepare cleaned gene sets for Venn plotting

Usage

gexp_qc_prepare_venn_sets(all_genes_list, max_sets = 5L)

Arguments

all_genes_list	Named list of gene vectors by dataset.
max_sets	Maximum number of sets to keep for Venn plotting.

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

---

Rebuild per-dataset gene lists from expression/count lists

Description

Rebuild per-dataset gene lists from expression/count lists

Usage

gexp_rebuild_all_genes_list(micro_expr_list, rna_counts_list)

Arguments

micro_expr_list	Named list of microarray matrices.
rna_counts_list	Named list of RNA-seq matrices.

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))

---

Run differential expression analysis

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

expr	Matrix of (batch-corrected, normalized) expression values with genes in rows and samples in columns.
metadata	Data.frame with at least a Condition column.
method	Currently only "limma" is implemented.
logfc_cutoff	Numeric log2 fold-change cutoff.
padj_cutoff	Numeric adjusted P-value cutoff.

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)

---

Prepare expression and sample data for WGCNA

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

expr	Matrix of expression values (genes x samples), typically batch-corrected.
metadata	Data.frame with sample metadata; should contain either rownames matching colnames(expr) or a SampleID column.
gene_mode	"all_common" or "top_variable" (default).
top_genes	If gene_mode = "top_variable", the number of most variable genes to select (default 5000).
min_samples_frac	Minimum fraction of samples with non-missing expression required for a gene to be kept (default 0.5).

Value

A list with elements:

datExpr	numeric matrix (samples x genes) for WGCNA
sample_info	data.frame with sample metadata aligned to datExpr
gene_variance_table	data.frame with gene variance and rank

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)

---

Build reproducibility report text for export

Description

Build reproducibility report text for export

Usage

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

Arguments

params	Named list of scalar analysis parameters (character or numeric).
include_session	Include sessionInfo() block (default TRUE).

Value

Character vector of report lines.

Examples

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

---

Summarise Dataset x Condition confounding for batch/DE guidance

Description

Summarise Dataset x Condition confounding for batch/DE guidance

Usage

gexpipe_batch_confounding_summary(metadata)

Arguments

metadata

data.frame with Dataset and Condition columns.

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)

---

Summarise how Platform should enter batch/DE models

Description

Summarise how Platform should enter batch/DE models

Usage

gexpipe_batch_covariate_info(metadata)

Arguments

metadata

sample metadata data.frame.

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)

---

Build ComBat / removeBatchEffect model matrix (biology to preserve)

Description

Build ComBat / removeBatchEffect model matrix (biology to preserve)

Usage

gexpipe_build_batch_mod(metadata)

Arguments

metadata

sample metadata data.frame.

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)

---

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

Description

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

Usage

gexpipe_build_de_design(metadata)

Arguments

metadata

data.frame with Dataset, Platform, Condition columns.

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)

---

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

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

meta_used	Metadata rows actually used in the DE fit.
total_meta	Full unified metadata before subsetting (optional).
method	DE method name.

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)

---

DESeq2 design formula for count-based DE

Description

DESeq2 design formula for count-based DE

Usage

gexpipe_deseq2_design(metadata)

Arguments

metadata

sample metadata data.frame.

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)

---

Detect microarray + RNA-seq in the same analysis

Description

Detect microarray + RNA-seq in the same analysis

Usage

gexpipe_has_mixed_platforms(metadata)

Arguments

metadata

data.frame with optional Platform column.

Value

logical

Examples

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

---

Independent filtering for DE (limma filterByExpr)

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

expr	Numeric matrix (genes x samples) or integer counts.
design	Model matrix for the DE analysis.
group	Optional factor when design is NULL (single-factor designs).

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)

---

Polar PCA coordinates for batch/platform diagnostic plots

Description

Polar PCA coordinates for batch/platform diagnostic plots

Usage

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

Arguments

expr	genes x samples matrix.
metadata	sample metadata aligned to expr columns.
color_by	column name in metadata to colour points.

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)

---

Test whether Platform is not estimable alongside Dataset

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

metadata

data.frame with Dataset and Platform columns.

Value

logical

Examples

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

---

Simplified PVCA variance bar-chart data (PCA + per-factor R^2 on top PCs)

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

expr	Numeric matrix, genes x samples.
metadata	Sample metadata with Dataset and optional Platform, Condition.
max_samples	Subsample at most this many columns for speed (default 100).
max_pcs	Use at most this many principal components (default 10).

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)

---

Set up GExPipe dependencies and optionally launch the app

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

update	Logical. If TRUE (default), update already-installed packages when auto-install is enabled.
optional	Logical. Reserved for future optional feature packages. Currently all runtime dependencies are in DESCRIPTION Imports and are always installed regardless of this flag.
launch	Logical. If TRUE, launch the GExPipe Shiny app after setup completes. Default FALSE.
port	Integer. Port for the Shiny app when launch = TRUE.
...	Additional arguments passed to shiny::runApp() when launch = TRUE.

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)
}

---

Drop redundant combined trait column from WGCNA module-trait correlation matrix

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

cor_mat	Numeric matrix of module-trait correlations.
combined	Name of the combined contrast column to drop, or NULL.

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")

---

Run the GExPipe Shiny application

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

launch.browser	If TRUE, open the app in the default browser (default). Set to FALSE in Google Colab or headless servers (no local browser).
port	TCP port for Shiny. Use a positive port (e.g. 3838) when you will open the app manually in a browser. 0 means "pick a free port" (fine for automated tests); some setups print ⁠http://127.0.0.1:0⁠, which cannot be pasted into a browser - then use launch.browser = TRUE or port = 3838.
host	The host to bind. Use "0.0.0.0" in Google Colab so port forwarding works.

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, port = 3838L)  # Step 2: start the server
}

---