Interactive web reports and static PDF reports

[matthiaskoenig]

The goal of this GSoC project is to design and implement an automated reporting framework for systems biology modeling workflows based on SBML, SED-ML, and PETab integrated into sbmlsim.

The project will:

• Generate interactive web reports using Quarto (interactive figures and tables).
• Generate static PDF reports using Typst (publication-ready summaries and figures).
• Integrate report generation into CI/CD pipelines (GitHub Actions).
• Automatically publish web reports via GitHub Pages (github.io).

The final outcome will be reusable reporting modules and templates within sbmlsim that can be adopted across multiple systems biology and systems medicine projects.

• sensitivity report
• create report (quarto); interactive images
• report for supplement => PDF; This should also contain documentation of the methods;
• Automatic generation of supplements and reports for publications
• models
• simulations
• parameter optimization
• literature review (PRISMA)

This will be addressed as part of GSOC nrnb/GoogleSummerOfCode#281

[kushvinth]

Hi @matthiaskoenig

This looks like a great direction for sbmlsim and very valuable for reproducible modeling workflows .

I’d be interested in contributing to this (as part of the GSoC effort if appropriate). Here’s how I’d approach the implementation:

Phase 1: Reporting Foundations

• Define a modular report architecture (SimulationReport, SensitivityReport, OptimizationReport) driven purely from model outputs and metadata.

• Create Quarto templates for interactive web reports (figures, tables, parameter summaries, diagnostics).

• Create Typst templates for publication-ready PDF supplements including:

  • methods documentation
  • model equations / parameter tables
  • simulation and optimization summaries

Phase 2: Pipeline Integration

• Add report-generation hooks inside sbmlsim workflows so reports can be produced automatically after simulations/optimizations.

• Provide GitHub Actions examples for:

  • generating Quarto web reports
  • building Typst PDFs
  • uploading artifacts and publishing to GitHub Pages

Phase 3: Extended Automation

• Support automatic generation of publication supplements combining:

  • models (SBML)
  • simulations (SED-ML)
  • parameter estimation results (PETab)
  • sensitivity analyses

• Optional additions:

  • PRISMA-style diagrams for literature workflows
  • provenance tracking linking model → simulation → results → report

Design principles I’d follow

1. Reproducible reports: fully generated from simulation outputs + metadata
2. Template-based + extensible: base templates with domain-specific extensions
3. Metadata-driven documentation: leverage existing annotations to auto-fill report sections

Happy to refine this based on maintainers’ priorities or existing design discussions. Please let me know if this aligns with the intended direction before I start exploring a draft implementation or PR.

[qzMalekuz]

Hi @matthiaskoenig ,

I’ve been reviewing the current sensitivity example workflow and how results are generated (TSV, PKL, PNG, TEX files per method). It seems that reporting is currently file-oriented, but not yet consolidated into a structured, publication-ready document.

For #180 specifically (interactive web + static PDF reports), I’m thinking a practical starting point could be:

Prototyping a minimal Quarto-based sensitivity report (e.g., for SamplingSensitivityAnalysis) that:

Aggregates parameter tables and compact sensitivity results,

Embeds generated figures,

Documents analysis settings (N, seed, bounds, method),

Produces a single self-contained HTML report.

In parallel, a corresponding Typst template could generate a PDF supplement that includes:

Methods description (sampling strategy, parameter ranges),

Parameter tables,

Sensitivity summaries and figures,

Model and simulation metadata.

My goal would be to keep this incremental and reusable, so the same structure can later support optimization and simulation reports.

Before drafting a prototype, would you prefer starting with:

A single concrete report type (e.g., sensitivity only), or

A small base reporting structure that can later be extended to optimization and simulation workflows?

Happy to align with your preferred scope.

Best,
Zafar.