Skip to content

Day 2 — Tuesday Jun 24, 2025

Day 1 (Mon Jun 23) was arrival-only — registration, dinner, an opening ceremony, and the "GBCC2025 LIVE!" evening session — so the program proper begins on Tuesday. Day 2 is the densest day of the conference and the one that establishes the joint Galaxy + Bioconductor character: an opening keynote on browser-side genomic computation and AI-assisted analysis, a Galaxy-infrastructure-heavy oral session built around AI agents and automatic R-package wrapping, five concurrent hands-on workshops at lunch, and then two parallel oral sessions in the afternoon — 2A on Galaxy workflows / ML / multi-omics, 2B almost entirely Bioconductor packages — totaling 24 talks before the wine-and-cheese and the first poster session.

Morning

Sergei L. Kosakovsky Pond (Temple) opened with Interactive Genomic Analytics with responsive notebooks, in-the-browser computation, and AI-assisted development. The framing — notebooks that run in the browser plus AI-assisted authoring — set up the first oral session almost too perfectly: Pond's "responsive, in-browser, AI-augmented" triad recurs as a structural argument across the four following talks.

Oral Session 1 (chaired by Sehyun Oh) leaned hard into AI-augmented Galaxy. GAIA (Junhao Qiu, Jeremy Goecks) opened with a natural-language interface that orchestrates analysis across Galaxy instances — the unified-NLI framing makes this a flagship Galaxy-AI artifact for 2025. GalaxyMCP (Dannon Baker, Qiu, Goecks, Mike Schatz) followed with the Model Context Protocol implementation that lets external AI assistants drive Galaxy programmatically — together with GAIA, this is the "AI in workflows" axis of the meeting. The third talk, bioc-to-galaxy (Jayadev Joshi, Fabio Cumbo, Daniel Blankenberg), was the integration-scaffolding centerpiece: an automated pipeline that wraps R/Bioconductor tools as Galaxy tools — the load-bearing engineering behind the rest of the joint conference's "Bioc-in-Galaxy" thread.

The session closed with three platform talks (Marius van den Beek on YAML-authored Galaxy tools; Aysam Guerler et al. on a new Galaxy visualization framework; Vincent Carey on language-agnostic genomic data representation in Bioconductor — a framework talk rather than a package launch), a materials-science domain-transfer case study (Liborio at STFC), an image-analysis-in-Galaxy talk (Kostrykin/Serrano-Solano), and AnVIL (Stephen Mosher, Schatz, Lawson, Carroll) — the NHGRI cloud platform as the at-scale endpoint for the whole stack.

Lunch ran five concurrent workshops: Marius van den Beek on Galaxy workflows; Stephen Salerno on inference for AI-predicted outcomes (a methodological piece that reappears as a Day 4 oral); John Davis / Ahmed Awan on Galaxy backend architecture; Greg Watson on autonomous workflows for materials laboratories; and Ellis Patrick on spatial-omics analysis strategies — the Patrick workshop is the spatial-Bioc anchor for Day 2 and pairs naturally with the sosta talk on Day 3.

Afternoon

The afternoon split into two parallel oral sessions — Galaxy-and-ML in 2A, almost-entirely-Bioconductor-packages in 2B — running 12 talks each.

Session 2A (Bush Hall, chaired by Jenny Drnevich) focused on Galaxy interfaces, ML, and integration. The infrastructure block covered a unified Galaxy run interface (Awan/Baker), more-powerful dataset collections (John Chilton, van den Beek, Awan), and AI-augmented analyses inside Galaxy (Qiu/Goecks — likely a GAIA-adjacent extension). The ML block included quantum-ML benchmarking (Raubenolt/Mohan), a deep-learning-with-torch epigenetic-clock case study (Jamie Park), and the Galaxy-ML2 tool suite (Lyra Junior, Qiu) — the ML-best-practices toolkit. Cross-domain Galaxy then surfaced with NOVA (Watson) for neutron-scattering data analysis. The integration block closed the session: anvi'o auto-wrapped via the same pipeline (bioc-to-galaxy, Cumbo/Joshi/Blankenberg); TPV Broker (De Geest, Srikakulam) for dynamic meta-scheduling; a single-cell case study on cardiomyocytes-in-AF (Loach/Bacon); gINTomics (Velle, Patanè, Romualdi) — multi-omics integration as a Bioconductor package — and Bioconductor-R-Shiny applications running on Galaxy (Videm/Zierep), which is conceptually the inverse of the auto-wrapping pipeline.

Session 2B (Grace Auditorium, chaired by Vincent Carey) was the first heavily-Bioconductor block of the meeting — six R packages presented back-to-back, plus four methods talks and two Galaxy pipelines. The package block: RAIDS (Pascal Belleau, Xintong Li, Astrid Deschênes) for ancestry inference from challenging molecular data; signifinder (Stefania Pirrotta, Laura Masatti) — public gene-expression cancer signatures across bulk, single-cell, and spatial; SEMplR (Grace Kenney, Sherpa, Boyle, Phanstiel) for predicting genetic-variant effects on TF binding. Methods talks without a single package anchor in the title: Kasper Hansen on allele-specific methylation via ONT long-reads; Pariksheet Nanda / Jason Shoemaker on host-cell signaling network perturbation; Xiuwen Zheng / Raven Chen on whole-genome annotation with GDS files (Zheng's SeqArray/GDS lineage); Aidan Lakshman / Erik Wright on orthology at modern-genomics scale (Wright's DECIPHER/SynExtend lineage); Lambda Moses / Bianca Dumitrascu on multiscale lung-adenocarcinoma spatial-transcriptomics (Moses's Voyager / SpatialFeatureExperiment lineage); Beatriz Campillo / Michael Love on differential transcript usage. Pascal Belleau returned with a CLEC12A / neutrophil-activation case study. Two Galaxy pipelines closed the session: Meta-iPep (Katherine Do, Subina Mehta) for HLA-bound microbial-peptide immunopeptidogenomics, and HiFi-MAG (Tyler Collins, Alex Ostrovsky, Schatz) for long-read metagenomics with new profilers. The pairing of signifinder here with mitology on Day 4 — both Pirrotta-led — is one of the cleanest single-author through-lines of the meeting.

Posters / social / evening

5 PM wine-and-cheese networking, 6 PM dinner, then Poster Session 1 from 7:30 to 9:30 PM — roughly 25+ of the 50+-poster slate. The abstract book PDF was unreachable at fetch time, so the poster lineup is not enumerated here.

What to chase

  • The full lineup for Oral Session 2B's methods talks — Hansen, Zheng, Lakshman/Wright, Moses, Campillo/Love — pin which specific Bioconductor packages each talk launched or updated (likely SeqArray, SynExtend, Voyager/SpatialFeatureExperiment, fishpond, plus the methylation talk which may anchor on bsseq or methylKit).
  • The Pond keynote slides — the "in-browser computation" claim covers a wide design space (WebAssembly notebooks vs. browser-bundled bioinformatics toolchains); the recording will sharpen which technology stack he's promoting and how it intersects with GAIA / GalaxyMCP.
  • The two integration-scaffolding talks — bioc-to-galaxy (Joshi/Cumbo/Blankenberg) and Bioconductor-R-Shiny-on-Galaxy (Videm/Zierep) — together set up the Day 3 Blankenberg workshop and the post-conference GSVA→Galaxy CoFest deliverable; the dependency chain is worth tracing end-to-end.