Fine-mapping and TWAS analysis

This mini-protocol selects among univariate, multivariate, multigene, functional and summary-statistic fine-mapping routes according to the available data and analysis goal.

Miniprotocol Timing

Timing: TBD

Overview

The MNM regression module analyzes individual-level genotype and molecular-phenotype data. qtl_dataset_construct+susie_twas performs univariate SuSiE fine-mapping and estimates TWAS weights; mnm jointly analyzes multiple molecular traits with multivariate priors; mnm_genes extends the multivariate model across genes; and fsusie fine-maps functional or epigenomic phenotypes.

The RSS module instead combines GWAS summary statistics with an external LD reference. These five commands are alternative analysis routes rather than a mandatory chain.

Steps

Analysis goal	Commands to run, in order	Inputs
Univariate molecular-QTL fine-mapping and TWAS weights	1	Genotype BED set, phenotype manifest, covariates and association windows under input/colocboost/
Multivariate fine-mapping across molecular traits	2	Step 1 inputs plus input/colocboost/fine_mapping_meta.tsv
Multigene multivariate fine-mapping	3	Multivariate inputs plus phenotype-ID map and retained-sample list under input/colocboost/
Fine-map functional or epigenomic phenotypes	4	input/genotype/protocol_example.genotype.chr22.bed; phenotype manifest, covariates and windows
GWAS summary-statistic fine-mapping	5	input/rss_analysis/protocol_example.rss_mwe.gwas_meta.tsv; input/ld_reference/protocol_example.ld_meta_file.tsv

Run only the command matching the analysis goal.

1. Univariate fine-mapping and TWAS

What it does: qtl_dataset_construct+susie_twas builds the regional dataset, fits SuSiE and saves fine-mapping results and cross-validated TWAS weights.

sos run pipeline/mnm_regression.ipynb qtl_dataset_construct+susie_twas   --name protocol_example --cwd output/mnm/univariate   --genoFile input/colocboost/example.chr22.bed   --phenoFile input/colocboost/pheno_manifest_multicontext.tsv   --covFile input/colocboost/example_covariates.tsv   --customized-association-windows input/colocboost/association_windows.bed   --region-name ENSG00000130538 --transpose-covariates --save-data -j1

2. Multivariate fine-mapping

What it does: mnm jointly fine-maps multiple molecular traits using the analyses and prior settings specified by the fine-mapping metadata.

sos run pipeline/mnm_regression.ipynb mnm   --name protocol_example --cwd output/mnm/multivariate   --genoFile input/colocboost/example.chr22.bed   --phenoFile input/colocboost/pheno_manifest_multicontext.tsv   --covFile input/colocboost/example_covariates.tsv   --customized-association-windows input/colocboost/association_windows.bed   --fine-mapping-meta input/colocboost/fine_mapping_meta.tsv   --transpose-covariates --save-data -j1

3. Multigene multivariate fine-mapping

What it does: mnm_genes coordinates multivariate fine-mapping across genes while preserving phenotype identifiers and a common retained-sample set.

sos run pipeline/mnm_regression.ipynb mnm_genes   --name protocol_example --cwd output/mnm/multigene   --genoFile input/colocboost/example.chr22.bed   --phenoFile input/colocboost/pheno_manifest_multicontext.tsv   --covFile input/colocboost/example_covariates.tsv   --customized-association-windows input/colocboost/association_windows.bed   --pheno-id-map-file input/colocboost/pheno_id_map.tsv   --fine-mapping-meta input/colocboost/fine_mapping_meta.tsv   --keep-samples input/colocboost/keep_samples.txt --save-data -j1

4. Functional fine-mapping

What it does: fsusie applies functional SuSiE to epigenomic or other functional phenotypes and saves posterior fine-mapping results and residual data.

sos run pipeline/mnm_regression.ipynb fsusie   --name protocol_example --cwd output/mnm/fsusie   --genoFile input/genotype/protocol_example.genotype.chr22.bed   --phenoFile input/phenotype/protocol_example.pheno_manifest.tsv   --covFile input/covariate/protocol_example.covariates.tsv   --customized-association-windows input/finemapping/protocol_example.association_windows.bed   --cis-window 0 --max-cv-variants 5000 --save-data -j1

5. Summary-statistic fine-mapping

What it does: the RSS chain harmonizes regional GWAS summary statistics with the LD reference, runs SuSiE-RSS fine-mapping and produces a regional diagnostic plot.

sos run pipeline/rss_analysis.ipynb   generate_manifest+generate_gwas_sumstats+gwas_fine_mapping+gwas_rss_plot   --cwd output/mnm/rss --modular-script-dir code/script   --gwas-meta input/rss_analysis/protocol_example.rss_mwe.gwas_meta.tsv   --regions chr22:49355984-50799822   --ld-meta input/ld_reference/protocol_example.ld_meta_file.tsv

Output Files

Route	Output filename and relative path	Description
Univariate	output/mnm/univariate/**/*.univariate_bvsr.rds; output/mnm/univariate/**/*.twas_weights.rds	SuSiE posterior and TWAS-weight objects
Multivariate	output/mnm/multivariate/**/*.multivariate_bvsr.rds; output/mnm/multivariate/**/*.twas_weights.rds	Joint multivariate posterior and weights
Multigene	output/mnm/multigene/**/*	Per-region, per-gene multivariate results and metadata
Functional	output/mnm/fsusie/**/*.fsusie.rds; saved residual objects	Functional fine-mapping posterior and residual data
RSS	output/mnm/rss/fine_mapping/AD_Bellenguez_2022.chr22_49355984_50799822.gwas_finemap.rds; regional plot under output/mnm/rss/plots/	Summary-statistic fine-mapping result and diagnostic plot

Anticipated Results

Each route produces posterior inclusion probabilities and credible-set information for the selected region and analysis model. Routes that estimate prediction weights also save TWAS-weight objects. The RSS route additionally records the harmonized regional summary statistics and LD-based diagnostics.

Command interface

sos run pipeline/mnm_regression.ipynb -h

sos run pipeline/rss_analysis.ipynb -h