Workflows

Topic-driven worked examples showing the CTIR R packages in real use — each workflow tackles a concrete biomedical question end-to-end, with multiple CTIR packages composed into a single reproducible pipeline.

Each workflow is a standalone workflowr project — fully versioned and reproducible, rendered as a static site with code, results, and narrative side by side.

Curated demonstrations, not exhaustive documentation. Clone, adapt, extend.

Single-Cell Tissue Atlas — From Raw Counts to Annotated Atlas

Coming soon

Which cell populations and states characterize injured vs. healthy tissue, and how do they relate to spatial context?

End-to-end single-cell workflow on a synthetic injured-tissue dataset: QC, normalization, clustering, annotation, segmentation-derived spatial context, and a publication-ready cell-level report. Demonstrates the CTIR single-cell stack composed end-to-end.

Featured packages: cellreportR, scimagR, segmantR, bambamR, songR

What you'll learn
  • Scaffolding a reproducible single-cell project with workflowr
  • QC, normalization, and SONG-based embeddings (songR) for interpretable structure
  • Linking segmentation (segmantR) and cell imaging (scimagR) outputs to expression data
  • Producing publication-ready cell-level reports with cellreportR
  • Cross-referencing bulk reference profiles via bambamR for population calibration

Hyperspectral Tissue Characterization — Imaging to Tissue Class Maps

Coming soon

Can tissue regions in injured tissue be classified from hyperspectral signatures alone, and how interpretable is the resulting embedding?

End-to-end HSI workflow: raw cube ingest, preprocessing, spectral library matching, dimensionality reduction, and tissue-region classification on a synthetic hyperspectral dataset. Composed from CTIR's HSI and dimensionality-reduction packages.

Featured packages: hyperspectR, libscanR, songR, reflowR

What you'll learn
  • Ingesting and preprocessing HSI cubes with hyperspectR
  • Reference-spectrum matching via libscanR
  • SONG-based interpretable embeddings (songR) over spectral features
  • Tissue-region classification and map generation
  • Building a CTIR-styled reproducible report site with reflowR

Multiplex Protein Quantification — From Plate Images to Cohort Inference

Coming soon

How do circulating protein biomarker panels stratify outcome groups in a synthetic trauma-like cohort?

Wet-lab-to-statistics workflow: parse .Q-View chemiluminescent multiplex ELISA project files, run plate-level quantification and QC, then perform cohort-level statistical analysis and reporting. Connects assay-level processing to inference.

Featured packages: qviewparsR, pressR, songR

What you'll learn
  • Parsing the binary .Q-View format with qviewparsR
  • Standard curve fitting, plate QC, and outlier handling
  • Per-plate and cohort-level reporting with pressR
  • Cohort stratification and exploratory embeddings with songR
  • Building a reproducible analysis trail from raw plate to inference

Bulk Transcriptomic Response — DE, Pathways, and Signature Discovery

Coming soon

Which transcriptional programs separate injury severity groups, and which signatures are robust across subcohorts?

Bulk RNA-seq workflow from count matrices to differential expression, pathway enrichment, and signature discovery, with reproducible figures and tables. Combines CTIR's bulk RNA-seq toolkit with interpretable embeddings.

Featured packages: bambamR, songR, reflowR

What you'll learn
  • Running bambamR from counts to DE and pathway results
  • Exploratory and signature-level embeddings with songR
  • Subcohort robustness checks and sensitivity analyses
  • Producing publication-ready figures, tables, and a reproducible report

Molecular Pathology Cohort — From Heterogeneous Reports to Queryable Cohort

Coming soon

How can heterogeneous molecular pathology data (NGS files, structured reports, free-text PDFs, clinical metadata) be unified into a queryable cohort and explored interactively?

Workflow for ingesting VCF/BAM/FASTQ, XML and PDF reports, and clinical metadata into a unified queryable database, with cohort-level summaries and an interactive Shiny exploration layer. Pairs ingestion with downstream cohort statistics.

Featured packages: molpathR, songR, pressR

What you'll learn
  • Ingesting heterogeneous molecular pathology files with molpathR
  • Building a queryable cohort database with provenance
  • Cohort-level descriptive statistics and embeddings via songR
  • Standardized cohort reports with pressR
  • Launching the interactive Shiny exploration app

Multi-Modal Tissue Integration — HSI x Single-Cell x Bulk

Coming soon

Can hyperspectral tissue maps, single-cell composition, and bulk transcriptomic signatures be jointly aligned to produce a coherent multi-modal view of injured tissue?

Cross-modality integration workflow: align HSI tissue-class maps, single-cell composition estimates, and bulk RNA-seq signatures across matched synthetic samples, with shared embeddings and consistency checks. The flagship 'puts it all together' workflow.

Featured packages: hyperspectR, cellreportR, bambamR, songR, segmantR

What you'll learn
  • Harmonizing modalities at the sample/region level
  • Joint embeddings with songR across modalities
  • Cross-modality consistency and disagreement diagnostics
  • Composite reporting that surfaces concordance and conflict
  • Limits and pitfalls of multi-modal alignment with synthetic data

Method Benchmarking — Synthetic Ground Truth and Recovery

Coming soon

Under what conditions do CTIR analysis methods recover known signal, and how do they compare to standard alternatives?

Benchmarking workflow that generates synthetic datasets with known ground truth, runs matched analyses across CTIR and reference methods, and quantifies recovery, calibration, and failure modes. Designed as a methods-paper companion template.

Featured packages: dynasimR, songR, bambamR, reflowR

What you'll learn
  • Configuring synthetic scenarios with dynasimR (effect sizes, noise, confounding)
  • Running matched analyses across methods on identical inputs
  • Quantifying recovery, calibration, and false-discovery behavior
  • Producing a benchmark report site with reflowR themes
  • Reusing the template for new methods papers

Templating a CTIR-Styled Reproducible Site

Coming soon

How do I scaffold a new CTIR-styled reproducible analysis site from scratch, with consistent theming and branding?

Meta-workflow for scaffolding a new workflowr project with CTIR theming, branded hex-sticker logo, navigation, and reproducibility checks — usable as the starting point for any of the workflows above.

Featured packages: reflowR, hexmakR

What you'll learn
  • Initializing a workflowr project with reflowR themes
  • Generating a hex-sticker logo with hexmakR
  • Structuring analysis pages, navigation, and reproducibility checks
  • Publishing to GitHub Pages with consistent CTIR branding