Differential Interactome Meta-Analysis Pipeline

The pipeline identifies rewired proteins — proteins that undergo significant changes in their interaction landscape without proportional changes in gene expression — across eight independent RNA-seq studies comparing Fusarium response to fungal and bacterial biological control agents (BCAs).

---

Methodology Reference

The Differential Interactome methodology (Q-value framework) used to quantify protein-protein interaction rewiring is adapted from:

Gulfidan, G., Turanli, B., Beklen, H. et al. (2020). "Pan-cancer mapping of differential protein-protein interactions." Scientific Reports 10, 3272. https://doi.org/10.1038/s41598-020-60127-x

Protein-protein interaction data: STRING v12.0 (combined score ≥ 400, taxid 229533 for F. graminearum).

---

Repository Structure

scripts/
├── 00_setup.R                  # Shared configuration (sourced by all scripts)
├── 01_orthology_mapping.R      # Cross-species gene ID mapping via KEGG KO
├── 02_differential_interactome.R  # Core Q-value differential PPI algorithm
├── 03_rewired_analysis.R       # Rewired protein identification and classification
├── 04_deg_volcano.R            # Volcano plots with rewired protein overlay
├── 05_overlap_analysis.R       # Venn/UpSet cross-study overlap analysis
├── 06_enrichment.R             # KEGG + GO hypergeometric enrichment
└── 07_network_module.R         # Network construction, Louvain modules, hub genes

---

Script Descriptions

00_setup.R — Shared Configuration

Sourced at the top of every pipeline script. Defines all file paths, analysis thresholds, study metadata (8 studies × organism/BCA/condition), shared color palettes, and utility functions (save_fig(), ensure_dir()).

No inputs required. Sets BASE_DIR relative to the script location automatically.

---

01_orthology_mapping.R — Cross-Species Orthology

• Input: KEGG KO assignment files (fgr_ko.txt, fox_ko.txt, fpu_ko.txt)
• Process: Maps F. oxysporum (FOXG) and F. pseudograminearum (FPSE) gene identifiers to F. graminearum (FGSG) IDs via shared KEGG Orthology groups. One-to-many mappings are resolved deterministically by selecting the lowest FGSG number.
• Output: resources/fox_to_fgr_orthology.csv, resources/fpu_to_fgr_orthology.csv

---

02_differential_interactome.R — Core Q-value Algorithm

• Input: featureCounts expression matrices (counts/1.txt … counts/8.txt), STRING PPI network, orthology mappings from Step 01
• Process: For each STRING edge in each study:
  1. Binarizes gene expression per sample (1 if expression > mean − 1 SD, else 0)
  2. Counts co-expression in Control (N_NC) and Treatment (N_ND) conditions
  3. Computes Q-value = N_ND / (N_NC + N_ND)
  4. Classifies edges: Activated (Q ≥ 0.90), Repressed (Q ≤ 0.10)
  5. Applies minimum frequency filter: max(N_NC, N_ND) ≥ Δ = 0.20
• Output: Per-study protein_stats.csv and interaction_stats.csv under outputs/differential_interactome/Study_N/

---

03_rewired_analysis.R — Rewired Protein Identification

• Input: protein_stats.csv from Step 02, DEG lists (degs/)
• Process: A protein is classified as rewired if it meets both criteria:
  1. Differential degree (N_total) in the top 25% across the study
  2. Not a DEG: |log₂FC| ≤ 1.0 or P_adj ≥ 0.05
• Output: rewired_proteins.csv per study under outputs/rewired_analysis/Study_N/; aggregated outputs/tables/rewired_summary.csv

---

04_deg_volcano.R — Volcano Plots (optional)

• Input: DEG lists, rewired protein lists from Step 03
• Process: Generates volcano plots (log₂FC vs −log₁₀ P_adj) for each of the 8 studies. Rewired proteins are overlaid as highlighted points; top-ranked rewired proteins are labeled.
• Output: outputs/figures/volcano_*.png/.svg; panel figure Fig1_volcano_panel_all_studies

---

05_overlap_analysis.R — Cross-Study Overlap (optional)

• Input: Rewired protein lists from Step 03, DEG lists
• Process: Computes pairwise and multi-way overlaps. Generates four-way Venn diagrams (fungal BCAs; bacterial BCAs separately) and an UpSet plot for all 8 studies combined.
• Output: outputs/figures/FigS2_venn_fungal_rewired, FigS3_venn_bacterial_rewired, Fig3_upset_rewired_proteins

---

06_enrichment.R — Functional Enrichment

• Input: Rewired protein lists from Step 03; KEGG KO file (fgr_ko.txt); UniProt GO annotations (fgr_uniprot_go.tsv); STRING alias table for ID mapping
• Process: Hypergeometric enrichment test for KEGG pathways and GO terms (MF/CC/BP). P-value computed as P(X ≥ k) = phyper(k − 1, m, n, N, lower.tail = FALSE). FDR correction by Benjamini–Hochberg. Pathway names fetched live from the KEGG REST API (rest.kegg.jp).
• Output: Enrichment tables and dot plots under outputs/enrichment/ and outputs/figures/Fig6_kegg_*, Fig7_go_*

---

07_network_module.R — Network Construction and Modules

• Input: Rewired protein lists from Step 03, STRING PPI network
• Process:
  1. Builds aggregated PPI sub-networks for fungal and bacterial rewired proteins
  2. Applies Louvain community detection (igraph::cluster_louvain(), seed = 42, default resolution = 1)
  3. Annotates modules with functional labels and hub genes (highest intra-module degree)
  4. Saves node/edge/hub CSVs for downstream network visualization
• Output: Network CSVs under outputs/networks/; module table outputs/tables/module_summary.csv; figures Fig4a/b, Fig5_module_traits_heatmap

---

How to Run

Scripts are designed to be run sequentially. Each script sources 00_setup.R automatically.

# Run full pipeline in order
Rscript 01_orthology_mapping.R
Rscript 02_differential_interactome.R
Rscript 03_rewired_analysis.R
Rscript 04_deg_volcano.R       # optional
Rscript 05_overlap_analysis.R  # optional
Rscript 06_enrichment.R
Rscript 07_network_module.R

Note: Scripts 04 and 05 are optional visualization scripts and do not affect downstream analysis.

---

Required Input Data

project_root/
├── counts/
│   └── 1.txt … 8.txt              # featureCounts output (tab-separated, with header)
├── degs/
│   └── {study_id}_{Up,Down}regulated_Genes.xlsx
├── metadata/
│   ├── study_metadata.csv          # Sample-level metadata (study_id, condition, replicate)
│   └── study_table.csv             # Study-level metadata (organism, BCA, type)
├── string_data/
│   ├── 229533.protein.links.v12.0.txt   # F. graminearum STRING PPI
│   ├── 229533.protein.aliases.v12.0.txt # F. graminearum protein aliases
│   ├── 426428.protein.aliases.v12.0.txt # F. oxysporum protein aliases
│   ├── fgr_ko.txt                  # F. graminearum KEGG KO assignments
│   ├── fox_ko.txt                  # F. oxysporum KEGG KO assignments
│   ├── fpu_ko.txt                  # F. pseudograminearum KEGG KO assignments
│   └── fgr_uniprot_go.tsv          # F. graminearum UniProt GO annotations
├── resources/                      # Auto-generated by 01_orthology_mapping.R
└── scripts/                        # This pipeline

---

Required R Packages

install.packages(c(
  "dplyr", "readr", "readxl", "openxlsx", "tidyr", "tibble",
  "ggplot2", "ggrepel", "ggraph", "igraph",
  "pheatmap", "VennDiagram", "UpSetR", "RColorBrewer",
  "gridExtra", "svglite", "httr", "stringr", "futile.logger"
))

Optional: RCy3 for Cytoscape integration.

---

Key Thresholds

Parameter	Value	Source
Q-value — repression	≤ 0.10	Gulfidan et al. (2020)
Q-value — activation	≥ 0.90	Gulfidan et al. (2020)
Δ (minimum co-expression frequency)	≥ 0.20	Gulfidan et al. (2020)
DEG log₂FC threshold	> 1.0	Standard
DEG adjusted p-value	< 0.05	Standard
Rewired protein percentile	Top 25% of differential degree	This study
STRING PPI score	≥ 400 (medium confidence)	STRING v12.0

---

Output Structure

All outputs are written to outputs/ relative to the project root:

outputs/
├── differential_interactome/Study_N/   # Q-value tables per study
├── differential_expression/            # DEG summaries
├── rewired_analysis/Study_N/           # Rewired protein lists per study
├── enrichment/                         # KEGG + GO enrichment results
├── networks/                           # Node/edge/hub CSVs for network figures
├── modules/                            # Louvain module assignments
├── figures/                            # All generated figures (PNG + SVG)
└── tables/                             # Summary tables