Visualization and Integrated System for Transcriptomic Analysis • VISTA

Visualization and Integrated System for Transcriptomic Analysis

VISTA is a Bioconductor framework for RNA-seq differential expression that keeps counts, statistics, annotations, and figures in a single validated SummarizedExperiment-based object — so you can go from raw counts to a publication-ready narrative without rebuilding the same glue code each project.

BiocManager::install("VISTA")

📦 One object, one grammar — DESeq2, edgeR, limma-voom, and a DESeq2/edgeR consensus behind a single entry point.
🎨 40+ publication-ready plots — QC, differential expression, expression patterns, fold-change structure, enrichment, and deconvolution.
🧬 Bioconductor-native — extends SummarizedExperiment, so assay(), rowData(), colData(), and [ all work as expected.
📄 Reproducible reporting — export helpers plus a YAML-driven Quarto workflow.

Installation · Quick start · Why VISTA · Core workflow · Plot catalogue · Object design · Documentation · Citation

Installation

Release — from Bioconductor:

if (!requireNamespace("BiocManager", quietly = TRUE)) {
  install.packages("BiocManager")
}

BiocManager::install("VISTA")

Development version (GitHub):

if (!requireNamespace("pak", quietly = TRUE)) {
  install.packages("pak")
}

pak::pak("cparsania/VISTA")

Quick start

library(VISTA)

data("count_data", package = "VISTA")        # gene-by-sample counts
data("sample_metadata", package = "VISTA")   # sample annotations

prepared_counts  <- read_vista_counts(count_data, format = "matrix", gene_id_column = "gene_id")
prepared_samples <- read_vista_metadata(sample_metadata)
matched_inputs   <- match_vista_inputs(prepared_counts, prepared_samples)

vista <- create_vista(
  counts          = matched_inputs$counts,
  sample_info     = matched_inputs$sample_info,
  column_geneid   = matched_inputs$column_geneid,
  group_column    = "cond_long",
  group_numerator = "treatment1",
  group_denominator = "control",
  method          = "deseq2",
  log2fc_cutoff   = 1,
  pval_cutoff     = 0.05
)

comp <- names(comparisons(vista))[1]

vista                                                    # object summary
head(comparisons(vista)[[comp]][, c("gene_id", "log2fc", "padj")])

Already have a SummarizedExperiment? Skip the import step:

vista <- as_vista(se, group_column = "cond_long")

Explore it

# Quality control
get_pca_plot(vista, label = TRUE)
get_mds_plot(vista, use_group_colors = TRUE)
get_corr_heatmap(vista)

# Differential expression
get_volcano_plot(vista, sample_comparison = comp)
get_ma_plot(vista, sample_comparison = comp)
get_deg_count_barplot(vista)

# Expression and fold-change views
up_genes <- get_genes_by_regulation(
  vista, sample_comparisons = comp, regulation = "Up", top_n = 40
)[[comp]]

get_expression_heatmap(vista, genes = up_genes, kmeans_k = 3)
get_expression_barplot(vista, genes = up_genes[1:3], by = "sample", facet_by = "gene")
get_foldchange_heatmap(vista)
get_foldchange_lollipop(vista, sample_comparison = comp, genes = up_genes[1:6], facet_by = "gene")

Why VISTA

Most RNA-seq projects repeat the same sequence: normalize counts, fit models, extract contrasts, plot QC, label genes, summarize pathways, assemble figures. The friction is rarely the statistics — it is the glue code between them.

VISTA organizes that work around one validated object:

Component	Accessor
Normalized expression	`assay(x)` / `norm_counts(x)`
Raw filtered counts	`counts(x)`
Feature annotations	`rowData(x)`
Sample metadata	`colData(x)` / `sample_info(x)`
Differential expression tables	`comparisons(x)`
Analysis parameters	`cutoffs(x)`

Every accessor and plotting function reads that same object, so you move from raw counts to a consistent analysis narrative without switching data structures between steps.

Core workflow

1. Prepare counts and metadata

prepared_counts  <- read_vista_counts(count_data, format = "matrix", gene_id_column = "gene_id")
prepared_samples <- read_vista_metadata(sample_metadata)
matched_inputs   <- match_vista_inputs(prepared_counts, prepared_samples)

These helpers import plain matrices and data frames, featureCounts, STAR gene counts, HTSeq-count, tximport-like inputs, and RSEM gene results. No metadata sheet yet? Bootstrap one from the count sample names:

starter_metadata <- derive_vista_metadata(
  matched_inputs$counts,
  column_geneid = matched_inputs$column_geneid,
  parser = "auto"
)

2. Build the analysis object

vista <- create_vista(
  counts = matched_inputs$counts,
  sample_info = matched_inputs$sample_info,
  column_geneid = matched_inputs$column_geneid,
  group_column = "cond_long",
  group_numerator = "treatment1",
  group_denominator = "control",
  method = "limma"                       # or "deseq2", "edger", "both"
)

Use method = "both" for a DESeq2/edgeR consensus:

vista_consensus <- create_vista(
  ...,
  method = "both",
  result_source = "consensus"
)

vista_consensus <- set_de_source(vista_consensus, "edger")   # switch the active table

3. Adjust the model

# Sample-level covariates
vista_cov <- create_vista(..., covariates = "cell")

# Or a full design formula
vista_design <- create_vista(..., design_formula = ~ cell + cond_long)

4. Add feature annotations

vista <- set_rowdata(
  vista,
  orgdb = org.Hs.eg.db,
  columns = c("SYMBOL", "GENENAME", "ENTREZID")
)

5. Export and report

export_vista_assets(
  vista,
  out_dir = "vista_assets",
  include_data = c("comparison", "norm_counts", "sample_info")
)

file.copy(
  system.file("reports", "vista-report-template.yml", package = "VISTA"),
  "vista-report.yml"
)
run_vista_report("vista-report.yml")

Plot catalogue

Quality control and sample structure

get_pca_plot() · get_mds_plot() · get_umap_plot() · get_corr_heatmap() · get_pairwise_corr_plot()

Differential expression summaries

get_volcano_plot() · get_ma_plot() · get_deg_count_barplot() · get_deg_count_pieplot() · get_deg_count_donutplot() · get_deg_venn_diagram() · get_deg_alluvial()

Expression patterns

get_expression_heatmap() · get_expression_boxplot() · get_expression_violinplot() · get_expression_barplot() · get_expression_lollipop() · get_expression_scatter() · get_expression_lineplot() · get_expression_density() · get_expression_joyplot() · get_expression_raincloud() · get_expression_chromosome_plot() · get_expression_matrix()

Fold-change structure

get_foldchange_scatter() · get_foldchange_barplot() · get_foldchange_lollipop() · get_foldchange_boxplot() · get_foldchange_lineplot() · get_foldchange_heatmap() · get_foldchange_matrix() · get_foldchange_chromosome_plot()

Pathway and enrichment

get_msigdb_enrichment() · get_go_enrichment() · get_kegg_enrichment() · get_gsea() · get_enrichment_plot() · get_enrichment_chord() · get_pathway_genes() · get_pathway_heatmap()

Deconvolution (optional)

run_cell_deconvolution() · get_celltype_barplot() · get_celltype_group_dotplot() · get_celltype_heatmap()

Harmonized plot API

VISTA plotting functions share one argument grammar, so the same concept always uses the same name across plot families:

Argument	Controls
`sample_group`, `group_column`	Sample filtering and grouping
`sample_comparison` / `sample_comparisons`	One contrast / several contrasts
`genes`, `top_n`	Which features to show, and how many
`by`	Group-level vs sample-level view
`facet_by`	Layout by gene, group, comparison, or none
`sample_order`	Sample sequencing for per-sample plots
`display_id`	User-facing gene labels
`summarise`	Collapse replicates to group means
`color_by`, `palette`, `colors`	Colour control
`return_type`	`"plot"`, `"data"`, or `"both"`

Older argument names keep working and warn with the release in which they become defunct — see ?"VISTA-deprecated".

Bioconductor-compatible object design

VISTA extends SummarizedExperiment, so standard Bioconductor workflows apply.

# Standard Bioconductor access
assay(vista)[1:5, 1:5]
rowData(vista)
colData(vista)
metadata(vista)
vista[1:100, ]                      # subsetting keeps DE tables aligned

# VISTA accessors
comparisons(vista)
deg_summary(vista)
cutoffs(vista)
norm_counts(vista, summarise = TRUE)

# Validation
validate_vista(vista, level = "full")

Raw filtered counts are retained alongside the normalized assay, so an object can go straight back into DESeq2:

counts(vista)                                     # integer counts
dds <- as_deseq_dataset(vista, design = ~ cond_long)

Note counts(), as_deseq_dataset(), and [ are available in the development version and are scheduled for the next Bioconductor release.

Example analyses

Enrichment from a comparison

msig <- get_msigdb_enrichment(
  vista, sample_comparison = comp, regulation = "Up",
  orgdb = org.Hs.eg.db, species = "Homo sapiens", msigdb_category = "H"
)

go_bp <- get_go_enrichment(
  vista, sample_comparison = comp, regulation = "Up",
  ont = "BP", orgdb = org.Hs.eg.db, species = "Homo sapiens"
)

get_enrichment_plot(msig$enrich)

Consistent colour control

group_colors(vista)

vista <- set_vista_group_colors(
  vista,
  c(control = "#264653", treatment1 = "#E76F51")
)

vista_consensus <- set_vista_comparison_colors(
  vista_consensus,
  c(treatment1_VS_control = "#6C5CE7")
)

Documentation

📖 Package website · Function reference

Workflows

Article	Description
Complete RNA-seq workflow	End-to-end analysis of the `airway` dataset
DESeq2 vs edgeR	Comparing backends and building a consensus
Code economy	VISTA against a standard R workflow
Cell-type deconvolution	Optional deconvolution workflow

Visualization guides

Guide	Description
Preparing counts and metadata	Importing common count formats
Colour and palette design	Consistent colours across comparisons
Enrichment chord diagrams	Pathway–gene chord plots
Raincloud plots	Distribution views
Function reference	Narrative tour of the API

Local help works as usual:

?create_vista
?get_expression_heatmap
?get_go_enrichment
?run_vista_report

Citation

If you use VISTA in published work, cite the release used in your analysis:

citation("VISTA")

Parsania C (2026). VISTA: Visualization and Integrated System for Transcriptomic Analysis.
doi:10.18129/B9.bioc.VISTA, https://bioconductor.org/packages/VISTA/

Support

🐛 Report an issue
💬 Bioconductor Support

Contributing

Contributions should preserve reproducibility and backward compatibility. Before opening a pull request:

Add or update tests for functional changes.
Run devtools::document() if roxygen comments changed.
Run devtools::test() and R CMD check.
Update NEWS.md for user-visible changes.

License

GPL-3

VISTA

Contents