Cleaned analysis scripts (manuscript / reuse)

Reorganized versions of the main step scripts for the ASAP Multiome (PD substantia nigra) single-nucleus multiome pipeline. Each script keeps the original analysis logic for its core computational steps, adds a documented header, sources a shared config.R, consolidates library() calls, and removes package-install, interactive (tmux/debug), and non-English clutter.

Scope: Most scripts were further trimmed to core steps only — QC plots, summary tables, and downstream interpretation blocks (GO/GSEA, LocusZoom, motif analysis, cross-cell-type summaries, etc.) were removed so the folder is suitable as a manuscript supplement. 02-Filtering-Clustering.R still retains QC and UMAP figures because clustering and manual annotation depend on them. S-LDSC plotting remains in the original ../scripts/08-GWAS-LDSC-QTL-plot+Neurons.R (not copied here).

How to run

1. Edit config.R and set PROJECT_ROOT (and, if needed, CELLRANGER_DIR, MACS_PATH, LDSC_HOME, METADATA_XLSX) for your environment, or export the matching environment variables (ASAP_MULTIOME_ROOT, ASAP_CELLRANGER_DIR, ASAP_MACS_PATH, LDSC_HOME, ASAP_METADATA_XLSX).
2. Run an R script from this directory, e.g. Rscript 02-Filtering-Clustering.R. Every R script begins with source("config.R"), which sets the working directory to PROJECT_ROOT; all other paths are relative to it.
3. Shell scripts (08-GWAS-LDSC-QTL.sh, 11-bw-files-for-ATAC.sh) read the same environment variables and can be run with bash <script>.sh.

config.R centralizes output directories, canonical object paths (SEURAT_OBJECT, MACS_COUNTS), QTL/GWAS inputs (GENOTYPE_DIR, COVARIATES_FILE, SNP_LOC, GWAS_SUMSTATS, GWAS_CLUMPED), the cell-type list (CELL_TYPES), and diagnosis contrasts (COMPARISONS). Step-local run targets (e.g. ct <- "Neurons" in QTL preprocessing, CELL_TYPE <- "Microglia" in eRegulon GSEA) remain at the top of the relevant script — change them or wrap the script in a loop to process all cell types.

Environment setup

Verified on this project: the main analysis used R 4.3.3 / Python 3.12.3 (single_cell_env); SCENIC+ used a separate R 4.4.1 / Python 3.11.3 (scenicplus env). Do not mix these up — the submitted README’s “R 4.4.1 / Python 3.11.3” applies to SCENIC+ only.

Conda env	R	Python	Used for
asap-multiome (environment.yml)	4.3.3	3.12.3	Steps 01–11, 13 (Seurat, QTL, NEBULA, moloc, scMultiMap, …)
asap-scenicplus (environment-scenicplus.yml)	4.4.1	3.11.3	Step 12 SCENIC+ only
asap-peer (environment-peer.yml)	3.4.1	3.5.5	PEER blocks in steps 05/06 only

# Main pipeline
conda env create -f environment.yml
conda activate asap-multiome
# GitHub-only R packages (see header comments in environment.yml):
# nebula, moloc, scMultiMap

# SCENIC+ (step 12)
conda env create -f environment-scenicplus.yml
conda activate asap-scenicplus

# PEER (steps 05/06)
conda env create -f environment-peer.yml
conda activate asap-peer

File	Purpose
environment.yml	Main asap-multiome env (R 4.3.3, Python 3.12.3)
environment-scenicplus.yml	SCENIC+ asap-scenicplus env (Python 3.11.3 + pip stack)
environment-peer.yml	PEER asap-peer env (R 3.4.1, steps 05/06 only)

Install separately: Cell Ranger ARC, FastQTL, LDSC (LDSC_HOME in config.R).

Scripts (one per main pipeline step)

Script	Step	Summary
config.R	–	Shared paths, tool locations, and global parameters
01-Combining-Samples.R	01	Build per-sample multiome Seurat objects and merge
02-Filtering-Clustering.R	02	QC filtering, SCT/LSI, Harmony, WNN, doublets, annotation (QC/UMAP plots retained)
03-Callpeaks-CellTypes.R	03	MACS peaks, union peak set, counts, cell-type-specific peaks
05-caQTL-preprocess.R	05	Pseudobulk accessibility, PEER, FastQTL inputs (+ shell blocks)
06-eQTL-preprocess.R	06	Pseudobulk expression, PEER, FastQTL inputs (+ shell blocks)
07-moloc-coloc.R	07	moloc caQTL + eQTL + GWAS three-trait colocalization
08-GWAS-LDSC-QTL.sh	08	S-LDSC partitioned heritability (annotations, LD scores, h²)
09-PD-differential-nebula-deg.R	09	Single-cell NEBULA DEG across diagnosis contrasts
10-scMultiMap-peak-gene.R	10	scMultiMap peak–gene cis linking
11-bw-files-for-ATAC.sh	11	Per-cell-type normalized ATAC bigWig tracks
13-1-eRegulon-GSEA-core.R	13	eRegulon gene-set GSEA on NEBULA DEG rankings (per cell type)

Core-only trimming (what was removed)

Script	Removed (non-core)
03	Genomic-distribution barplot, UpSet, rGREAT GO, accessibility heatmap
05	QC plots, per-chromosome matrix dumps, raw pseudobulk writes, PEER alpha diagnostics
06	Same as 05 (retains one PEER alpha PDF for factor-count sanity check)
07	LocusZoom input prep, stacked locus plots, motif analyses
08	Repetitive shell blocks consolidated into a reusable run_sldsc function
09	GO over-representation and GSEA enrichment on DEGs
10	Cross-cell-type summary and moloc-intersection sections
13-1	Plotting helpers and diagnosis-specific RSS prefilter

Preserved throughout: analysis steps, statistical models, thresholds, and the primary output files each script writes.

Dependencies

Key R packages (via environment.yml; nebula/moloc/scMultiMap from GitHub — see yml header):

Step	Packages
01–03	Seurat, Signac, GenomicFeatures, EnsDb.Hsapiens.v86, dplyr, edgeR, harmony
05–06	edgeR, data.table, tidyverse; PEER via asap-peer env
07	moloc, data.table, GenomicRanges
09	nebula, RUVSeq, edgeR, Matrix
10	scMultiMap
13-1	data.table, clusterProfiler

See also the parent ../README.md for the full original pipeline dependency list.