MAXOMOD_CSF_Proteomics

This repository provides demo data and scripts to reproduce the bioinformatic pipelines and figures from the manuscript Proteomic Profiling of Cerebrospinal Fluid Identifies Immune and Synapto-Axonal ALS Subtypes.

---

Project structure

Path	Description
Script/	R scripts for the full analysis pipeline (00–20)
Analysis.sh	Master script: Discovery → Validation → between-cohort → External → Brain
vignettes/	R Markdown files to reproduce selected main/supplementary figures
demo/	Demo inputs (copy of pipeline outputs) used by vignettes
Discovery/, Validation/	Pipeline outputs for each cohort
Input/	Reference data (e.g. ID mapping, annotation files)
Plots/	Default figure output directory for vignettes

---

System requirements

R and RStudio

• R v.4.2.3 (or compatible). Install CRAN packages with install.packages(). For Bioconductor packages (e.g. DEP, SummarizedExperiment, clusterProfiler, org.Hs.eg.db) use BiocManager::install("PackageName").

Packages

• AnnotationDbi
• biomaRt
• boot
• caret
• circlize
• cluster
• clusterProfiler
• ComplexHeatmap
• cowplot
• data.table
• dendextend
• DEP
• dplyr
• dunn.test
• e1071
• GOSemSim
• ggalluvial
• ggpubr
• ggrepel
• ggsci
• ggplot2
• ggthemes
• glmnet
• grid
• gridExtra
• GSVA
• limma
• mclust
• msigdbr
• NMF
• nnet
• nortest
• openxlsx
• optparse
• org.Hs.eg.db
• patchwork
• pheatmap
• pROC
• purrr
• randomForest
• readr
• readxl
• reshape2
• RColorBrewer
• scales
• stats
• stringr
• SummarizedExperiment
• tibble
• tidyr
• tidyverse
• umap
• visdat
• WGCNA
• writexl
• xgboost

Operating System:

macOS Tahoe 26.2 (Apple Silicon, M3 Pro)

Any required non-standard hardware

None for the custom codes.

Instruction for use

Codes to generate all main and supplementary figures are in the Script/ folder. Run the full pipeline from the project root:

bash Analysis.sh

Or in the background: nohup bash Analysis.sh > Analysis.log 2>&1 &
Paths in scripts are relative to the project root; do not run from another directory.

Demo

Demo data are under demo/ (e.g. demo/Discovery/, demo/Validation/). Populate them by copying from pipeline outputs, or run the pipeline first. Then open and knit the R Markdown files from the project root (or from vignettes/; the documents set the project root automatically).

Figure	Vignette	Description
Fig. 1b	vignettes/Fig1b_clinical_heatmap.Rmd	Heatmap of the 100 most variable CSF protein abundances in the discovery cohort (run time: < 1 min).
Fig. 1c, 1d, 1e	vignettes/Fig1cde_volcano_scatter.Rmd	Discovery and validation volcano plots (1c, 1d) and Discovery vs Validation signed FDR scatter (1e). Uses demo/Discovery/ and demo/Validation/ DE results (run time: < 1 min).
Fig. 1f	vignettes/Fig1f_GESA_IC.Rmd	GSEA IC heatmap (GO terms clustered by semantic similarity) for Discovery and Validation. Uses demo/Discovery/06_GSEA/ and demo/Validation/06_GSEA/ (run time: < 1 min).
Fig. 2b, 2c	vignettes/Fig2bc_Clustering.Rmd	AIC by clustering method and number of clusters (2b); Sankey plot of cluster assignment from k=2 to k=3 (2c). Exports cluster_assignments_2.csv. Uses demo/Discovery/02_Missing_Inspection_subclusters/ (run time: < 1 min).
Fig. 2d, 2e, 2f	vignettes/Fig2def_volcano_scatter.Rmd	Subclusters (alpha vs beta): Discovery volcano (2d), Validation volcano (2e), Discovery vs Validation signed FDR scatter (2f). Uses demo/Discovery/ and demo/Validation/ subcluster DE results (03_Differential_expression_analysis_subclusters/) (run time: < 1 min).
Fig. 3a–3g	vignettes/Fig3_WGCNA.Rmd	WGCNA: dendrogram and module–trait bars (3a), eigengene networks (3b), top GO terms per module IC heatmap (3c), eigenprotein boxplots for turquoise and blue modules in Discovery/Validation/External (3d), module–trait heatmaps Discovery (3e) and Validation (3f), MEs vs clinical variables (3g). Saves PDFs to Plots/Fig3_WGCNA/. Uses demo/Discovery/02_Missing_Inspection_subclusters/, 08_Clustering_als/, and optionally Validation/External inputs (run time: ~ 8 min).

About issues:

To report bugs, ask questions, or provide feedback, please use the GitHub Issues page.