A 327-sample RNA-Seq study asking how Grapevine Red Blotch Virus rewires the Vitis vinifera (Merlot)
transcriptome, how temperature modulates that rewiring, and which host genes carry
the causal weight between the two.
flowchart TD
classDef env fill:#E69F00,stroke:#333,stroke-width:1px,color:#fff
classDef vir fill:#0072B2,stroke:#333,stroke-width:1px,color:#fff
classDef host fill:#8E44AD,stroke:#333,stroke-width:1px,color:#fff
classDef down fill:#2166AC,stroke:#333,stroke-width:1px,color:#fff
classDef phe fill:#B2182B,stroke:#333,stroke-width:1px,color:#fff
T["Hot season<br/>+7.6 degC at Harvest"]:::env
V["Viral titer<br/>biphasic<br/>3.6x early, 0.18x late"]:::vir
M["MBF1c<br/>HSF co-activator<br/>61.8 percent mediated"]:::host
S["29 of 36 sHSPs<br/>suppressed<br/>module score -55 percent"]:::down
I["Canonical immunity<br/>10 categories<br/>|log2FC| less than 0.08"]:::down
P["Disease phenotype<br/>red blotch<br/>sugar deficit"]:::phe
T -->|path a| V
V -->|path b| M
M --> S
T -. direct effect .-> S
V -. no trigger .-> I
S --> P
I -.->|silent| P
linkStyle 0 stroke:#E69F00,stroke-width:2px
linkStyle 1 stroke:#0072B2,stroke-width:2px
linkStyle 2 stroke:#8E44AD,stroke-width:2px
The headline is not that GRBV is fighting the plant.
It is that GRBV is blocking the plant's heat-induced defence programme from running —
a passive failure, not an active suppression — while slipping past canonical immune
defence without triggering it at all. Uninfected vines at the hottest harvest cell reach
sHSP module z = +1.64; infected vines at the same cell reach −0.67.
The 2.3-unit deficit is failure-to-activate, not active downregulation.
| Species | Vitis vinifera cv. Merlot, one commercial vineyard, 10+ blocks |
| Samples | 327 (264 RB+, 63 RB-) |
| Seasons | 2021 (cool), 2022 (hot, +7.6 °C at Harvest) |
| Maturity stages | Pre-véraison, Véraison, Post-véraison, Harvest |
| Infection cohorts | 7 onset years (2016–2022), n = 105–214 DEGs each |
| Sequencing | TAG-seq (UMI-indexed) · STAR · Pinot Noir ASM3070453v1 |
| Primary DE design | DESeq2 Wald, ~ Maturity_stage + Sampling_Season + Infection_Group, BH-FDR < 0.05, |log₂FC| ≥ 1 |
| Figures | 12 Main (incl. MoA schematic + graphical abstract) + 64 Supplementary (S02–S72), PNG + SVG |
| Presentation | 71-slide funding committee deck, 5 PLUS sections, 2 graphical-abstract slides |
Main Figure 3. The inverted seasonal DEG trajectory. 2021 peaks at Harvest (1,567 DEGs); 2022 peaks earlier at Post-véraison (772). Viral titer runs 3.6x higher in the hot year at Pre-véraison and 0.18x at Harvest: the biphasic titer pattern is the mechanistic hinge.
| 1 |
2021 peaks at Harvest with 1,567 DEGs. 2022 peaks earlier at Post-véraison with 772, falling back to 759 by Harvest. Cross-season fold-change correlation rises with maturity: r = 0.34 at Pre-véraison, r = 0.67 at Post-véraison. The direction switch is mechanistically driven by biphasic viral titer: higher early in hot years, lower late, flipping the timing of disease expression. |
| 2 |
29 of 36 small heat-shock proteins (sHSPs) are down-regulated in infected vines. HSF ↔ sHSP coordination stays coherent (r = 0.625, p = 7.2e-37) but the sHSP module score drops from +1.64 in healthy vines to +0.74 in infected (55 % reduction). All 10 canonical plant-immunity categories (NLR, PRR, PR, SA, JA, ET, WRKY, callose, ROS, R-adapters) show |log₂FC| < 0.08. GRBV is not triggering effector-triggered or pattern-triggered immunity. It is dampening thermal defense. |
| 3 |
Causal-mediation sweep on the top 2,000 variable genes returns 39 temperature → titer → gene mediators at joint BH-FDR < 0.05. The top hit, LOC100249249, turns out to be multiprotein-bridging factor 1c (MBF1c), a known HSF co-activator: joint p = 4.8 × 10⁻¹¹, 61.8 % of the temperature effect mediated by viral titer. Second-tier robust mediator is LOC100232974 = sucrose-phosphate synthase 1 (SPS1). Paired harvest biochemistry (Figs S61-S63, n = 33) shows SPS1 correlates negatively with Brix within-season (2021 ρ = −0.73), so SPS1 is up-regulated in the infected, low-Brix berries. The mediator status stands; the role is compensatory, not rate-limiting. |
| 4 |
Two-way Season × Infection ANOVA on n = 33 paired harvest samples confirms the transcriptomic inversion in the phenotype. TA is the cleanest mirror: no main effect of season (p = 0.69) or infection (p = 0.16), but the Season × Infection interaction reaches η² = 0.77, p = 4.3 × 10⁻¹¹. GRBV raises TA by +0.62 g/L in 2021 cool and lowers it by −0.79 g/L in 2022 hot; malic acid follows the same inversion (+398 vs −195 mg/L). Brix is suppressed in both seasons but 2.6× stronger in 2022 (−9.83 vs −3.77 °Bx). In 2021 a biphasic cohort-age recovery appears at harvest: Brix rises +0.10 °Bx / yr with years-since-infection (r = 0.76, p = 0.0045) while nitrogen markers decline. |
| 5 |
The disease progression cube (S66) puts RB(+) vines at sHSP module z = −0.67 at 2022 Harvest. The matched-condition RB(−) baseline cube (S67) puts uninfected vines at +1.64 in the same cell — the strongest natural heat-induced sHSP activation anywhere in the grid. The 2.3-unit deficit is not active downregulation from neutral; it is the vine failing to mount the normal heat response. The gene-level variance decomposition ternary (S72, methodologically after Singh et al. 2025 bioRxiv Fig 5) confirms it at the gene level: sHSP genes carry environmental variance share 0.33 (vs 0.19 baseline) but viral variance share 0.11 ≈ 0.12 baseline, indistinguishable from background. GRBV doesn't repress sHSPs — it prevents the cell from running the heat-induced HSF/MBF1c transactivation programme. The wet-lab target shifts accordingly. |
| 6 |
The 7-year cohort span (2016–2022 onset) separates into three phases. Acute (2022 cohort, year 1): 214 DEGs — the largest count — but the weakest pathway enrichment (≤ 5× fold for most curated families). Peak (2020 cohort, years 3–5): strongest PC-space drift from the matched RB(−) centroid (21.9 distance at 2021 Harvest, top-ranked cell across all 40 cube cells). Chronic (2016 cohort, years 6–7): only 105 DEGs but hyper-focused enrichment — 65× for sHSP, 138× for HSP100/ClpB, 49× for GO:response to heat. The GRBV 55-gene signature is the one family enriched 52–100× in every cohort as the persistent transcriptomic fingerprint. Pattern: acute diffuse → peak drift → chronic narrow. No prior GRBV study has had the cohort span to see this. |
| 7 |
RB(−) controls reach 26.7 °Bx in cool 2021 and 34.4 °Bx in hot 2022. Infected RB(+) vines reach 22.8–23.4 and 23.5–25.0 °Bx respectively. Disease cost: ≈ 4 °Bx in the cool year, ≈ 10 °Bx in the hot year — 2.6× amplification. This translates the transcriptional inverted-trajectory finding into a crop-value number a grower can act on: in a hot year, an infected vine costs the grower ten Brix points at harvest. |
flowchart LR
classDef old fill:#fce8e6,stroke:#B2182B,color:#5a0000
classDef fix fill:#fff4d8,stroke:#E69F00,color:#5a3d00
classDef new fill:#e3f0fb,stroke:#2166AC,color:#0a2b52
subgraph pre ["Pre-fix (Welch-t)"]
P1["908 universal DEGs<br/>pooled, uncorrected"]:::old
P2["AUC = 1.000<br/>no host/viral split"]:::old
P3["0 mediators<br/>at tv = 500"]:::old
P4["27/27 QTL overlap<br/>no null distribution"]:::old
end
subgraph fix ["Critical review: 13 fixes"]
F1["Covariate-adjusted DESeq2<br/>Stage + Season + Infection"]:::fix
F2["Leave-one-season-out CV<br/>host-only features"]:::fix
F3["Raise tv to 2,000"]:::fix
F4["10,000-permutation null<br/>chromosome-stratified"]:::fix
end
subgraph post ["Post-fix"]
Q1["126 core infection DEGs<br/>13.6 percent of the original"]:::new
Q2["Host LOSO AUC 0.73-0.88<br/>honest generalisation"]:::new
Q3["39 mediators<br/>MBF1c top p=4.8e-11"]:::new
Q4["fold = 2.01, p less than 0.0002<br/>defensible enrichment"]:::new
end
P1 --> F1 --> Q1
P2 --> F2 --> Q2
P3 --> F3 --> Q3
P4 --> F4 --> Q4
The data layer came out of the reanalysis richer (24 covariate-adjusted contrasts,
including per-stage x per-season cells). The manuscript claims came out honest.
Full documentation: CRITICAL_REVIEW.md, DE_REANALYSIS_REPORT.md, SENSITIVITY_REPORT.md, METHODS_ADDENDUM.md.
flowchart LR
classDef raw fill:#f4f4f4,stroke:#64748B,color:#0F172A
classDef pipe fill:#d6e3f5,stroke:#2166AC,color:#0a2b52
classDef out fill:#e3dcf0,stroke:#8E44AD,color:#2a164a
subgraph raw ["Raw data"]
direction TB
A1[TAG-seq reads<br/>UMI-indexed]:::raw
A2[Sample metadata<br/>vine_id, stage, season]:::raw
A3[Viral titer<br/>RT-qPCR + dPCR]:::raw
A4[Field weather<br/>daily 2021, 2022]:::raw
end
subgraph pipe ["HPC pipeline (Snakemake on FARM)"]
direction TB
B1[STAR + featureCounts<br/>Pinot Noir ASM3070453v1]:::pipe
B2[DESeq2<br/>24 stratified contrasts]:::pipe
B3[Causal mediation<br/>tv = 2,000]:::pipe
B4[WGCNA<br/>13 co-expression modules]:::pipe
B5[GSEA<br/>375 sets x 8 conditions]:::pipe
B6[QTL permutation<br/>10,000 chromosome-stratified]:::pipe
B7[NCBI annotation<br/>82 of 86 LOC resolved]:::pipe
end
subgraph out ["Deliverables"]
direction TB
C1[Figure atlas<br/>10 Main + 47 Supp<br/>PNG + SVG]:::out
C2[Manuscript draft<br/>docs/MANUSCRIPT_DRAFT_v1.pdf]:::out
C3[Funding deck<br/>48 slides, 19 MB]:::out
C4[Methods + sensitivity<br/>docs/]:::out
end
A1 --> B1 --> B2
A2 --> B2
A3 --> B3
A4 --> B3
B2 --> B3
B2 --> B4
B2 --> B5
B2 --> B6
B4 --> B7
B2 --> C1
B3 --> C1
B4 --> C1
B5 --> C1
C1 --> C2
C1 --> C3
B2 --> C4
Click to expand full tree
.
├── figures/
│ ├── main/ Main01-Main10 + MoA_GRBV_mechanism + Summary_study_design (PNG + SVG)
│ ├── supp/ S02-S72 supplementary figures (PNG + SVG, 64 figures)
│ ├── assets/icons/ 15 Firefly-generated PNG icons used by Main10, MoA, Summary
│ ├── scripts_R/ R / ggplot2 scripts for every figure (28 scripts)
│ │ └── theme_grbv.R Shared theme + palette + save_grbv() helper
│ ├── GRBV_Main_Figures_Atlas.pdf 10-page main atlas
│ ├── GRBV_Full_Figure_Atlas.pdf 75-page full atlas
│ └── UPGRADE_NOTES.md changelog + three-new-dimensions summary
├── farm_pipeline/
│ ├── Snakefile 13 rules, 24 DE contrasts, sensitivity sweeps
│ ├── scripts/
│ ├── results/ de_redo, de_extras, mediation, qtl, annotations + rules 00-12
│ ├── data/, resources/
│ └── docs/
│ └── LITERATURE_SCHEMA.md
├── docs/ Project-level deliverables (14 files)
│ ├── MANUSCRIPT_DRAFT_v1.{md,pdf}
│ ├── INTEGRATED_ANALYSIS.{md,pdf}
│ ├── CRITICAL_REVIEW.{md,pdf}
│ ├── DE_REANALYSIS_REPORT.md
│ ├── METHODS_ADDENDUM.md
│ ├── HSFA2_VIGS_EXPERIMENT_PLAN.md
│ ├── SENSITIVITY_REPORT.{md,pdf}
│ ├── EXTENDED_RESULTS_REPORT.{md,pdf}
│ └── PLAIN_LANGUAGE_SUMMARY.md
├── Old-Analysis/ Pre-fix era archive
│ ├── images/ 55 legacy PNGs from Python pipeline
│ ├── scripts_python/ 19 pre-fix analysis scripts
│ ├── generate_scripts/ 5 pre-fix PDF / PPTX builders
│ ├── reports/ 8 legacy deliverables
│ ├── superseded_csvs/ 11 CSVs with zero current-R-script references
│ └── farm_sensitivity_archive/ 58 MB raw threshold sweeps
├── GRBV_Funding_Committee_Presentation.pptx 71-slide committee deck
├── scripts/ Python PPTX updaters (add_ppt_new_analyses, refresh_ppt_figures, ...)
├── counts_matrix.txt + sample_metadata.txt + 20 active-pipeline CSVs
├── org.Vvinifera.eg.db/, go-basic.obo
├── CLAUDE.md Project memory / session context
└── README.md This file
Every figure is an R / ggplot2 script in figures/scripts_R/. Each script sources theme_grbv.R (project palette + save helper) and writes PNG + SVG to figures/main/ or figures/supp/.
# One figure
Rscript figures/scripts_R/main03_inverted_trajectory.R
# A whole supplementary batch (S02 through S07)
Rscript figures/scripts_R/supp_de_overview.R
# The new three-axis cube figures (S66, S67) + variance ternary (S72)
Rscript figures/scripts_R/S66_disease_progression_cube.R
Rscript figures/scripts_R/S67_rbminus_baseline_cube.R
Rscript figures/scripts_R/S72_variance_decomposition_ternary.R
# The two graphical-abstract Main figures (use icons from figures/assets/icons/)
Rscript figures/scripts_R/moa_grbv_mechanism.R
Rscript figures/scripts_R/summary_study_design.R
# Rebuild the atlases from the current PNG set
python figures/scripts_R/build_atlas.py
python figures/scripts_R/build_full_atlas.pyOutput formats
- PNG: 600 dpi, via
ragg::agg_png, Arial fonts reliable on macOS - SVG: vector, via
svglite::svglite, opens cleanly in Illustrator, Inkscape, Figma - PDF: opt-in with
save_grbv(..., formats = c("png","pdf"))viacairo_pdf - Atlases: 2 PDFs assembled directly from PNGs (reportlab)
The extended analysis runs as a Snakemake pipeline on UC Davis FARM.
cd farm_pipeline
bash setup_farm.sh # install R packages + annotation resources
bash run_farm.sh # submit to partition low, account publicgrp
bash pull.sh # rsync results back to local figures/The 71-slide deck is maintained by a set of in-place Python updater scripts in scripts/:
# Replace stale figures + insert Summary + MoA graphical-abstract slides
python scripts/refresh_ppt_figures.py
# Add the 4 "new-dimension" slides (passive failure, 3 phases, 2.6x cost)
python scripts/update_ppt_new_dimensions.py
# Add the 5 "new-analysis" PLUS slides (S29, S68, S69, S70, S72)
python scripts/add_ppt_new_analyses.pyEach script clones an existing 2-column template slide, swaps the image, overwrites the title + bullets, and positions the new slide in the correct section. Hash-audit confirms every embedded PNG matches the current figures/ folder.
| Asset | Location | Size |
|---|---|---|
| Manuscript draft (with figures embedded, ≈ 5,400-word running total) | docs/MANUSCRIPT_DRAFT_v1.pdf |
11 MB |
| Main figures atlas | figures/GRBV_Main_Figures_Atlas.pdf |
11 MB |
| Full atlas (main + supp) | figures/GRBV_Full_Figure_Atlas.pdf |
36 MB |
| Figure upgrade notes (incl. three-new-dimensions summary) | figures/UPGRADE_NOTES.md |
5 KB |
| Methods addendum | docs/METHODS_ADDENDUM.md |
16 KB |
| Committee deck (71 slides) | GRBV_Funding_Committee_Presentation.pptx |
30 MB |
| Wet-lab plan (VIGS) | docs/HSFA2_VIGS_EXPERIMENT_PLAN.md |
17 KB |
| Critical review + fix log | docs/CRITICAL_REVIEW.pdf |
23 KB |
| Integrated analysis (incl. Thread 10) | docs/INTEGRATED_ANALYSIS.pdf |
41 KB |
| Extended results report (incl. claims 10–12) | docs/EXTENDED_RESULTS_REPORT.pdf |
41 KB |
| Plain-language summary | docs/PLAIN_LANGUAGE_SUMMARY.md |
15 KB |
timeline
title Submission roadmap
section Validation
May-Jun 2026 : qPCR on 10 key genes : MBF1c, SPS1, 3 sHSPs, HSFA2, 3 biomarkers
Jun-Jul 2026 : GEO + SRA deposition : 14,604 genes x 327 samples
section Submission
Aug 2026 : Nature Plants submission : methods + data finalised
section Extension
Jul 2026 - Feb 2027 : VIGS on MBF1c + SPS1 : N. benthamiana then grapevine
Aug 2026 : Tier-1 meta-analysis : 5 published Vitis stress studies
| In memoriam | The founding PI, Dr. Anita Oberholster (UC Davis Viticulture and Enology, 1980 to 2025), passed away in January 2025. Her lab built the field-sampling design, ran the wet-lab, and framed the year-one manuscript. Every result here sits on that foundation. |
| Mentorship | Dr. David E. Block, UC Davis V and E. Current PI of record; manuscript strategy; resource allocation on FARM. |
| Virology co-authorship | Dr. Mysore R. Sudarshana, USDA-ARS and UC Davis Plant Pathology. Field-site liaison; viral-titer wet-lab; co-author on two in-preparation manuscripts. |
| RNA-Seq pipeline | Bradley N. Jenner, UC Davis Bioinformatics Core (2025-04-15 to 2025-05-12). TAG-seq QC, HTStream, umi_tools dedup, STAR, featureCounts. |
| Funding | USDA-ARS grapevine pathology programs; California Department of Food and Agriculture (CDFA) specialty-crop block grants. |
| Compute | UC Davis FARM HPC, partition low, account publicgrp. |
| Annotation | NCBI; org.Vvinifera.eg.db; Gene Ontology Consortium; KEGG. |
Singh, P.P., Block, D.E., Sudarshana, M.R., Oberholster, A. (in preparation).
Biphasic temperature drives viral titer dynamics and silences thermal defense in Vitis vinifera cv. Merlot under Grapevine Red Blotch Virus infection: a 327-sample transcriptomic study.
Contact: ppssingh@ucdavis.edu · ORCID 0000-0001-7921-9379 · prempsingh.com
MIT License · see LICENSE for details