All of Us RNA-seq eQTL and sQTL Analysis Pipeline

This section documents the All of Us RNA-seq QTL workflows for expression/splicing phenotype generation, QTL analysis preparation, and SuSiE fine-mapping aggregation.

Quick Summary

• Purpose: Run reproducible eQTL and sQTL analyses from genotype and RNA-derived inputs.
• Primary Outputs: RNA-derived phenotype artifacts, TensorQTL-ready inputs, SuSiE fine-mapping results, and aggregated annotations.

Input Requirements

To run the workflows described below, you generally need:

• A joint-called VCF containing the relevant samples
• Research IDs partitioned by ancestry or subpopulation
• RNA expression quantifications (for eQTL)
• BAM/CRAM files for splice junction extraction (for sQTL)
• Sample-level metadata tables

Ordered Analysis Flow

The table below reflects the original end-to-end run order, including steps that do not yet have dedicated docs pages in this folder.

Order	Stage	Workflow / Component	Documentation	WDL	Run next
0	Cohort setup	Ancestry grouping and sample lists	No dedicated WDL page (notebook/table prep)	N/A	Use ancestry/sample partitions as inputs to genotype and phenotype prep.
1	Genotype prep	Prepare genotypes (pruning + PLINK + PCs)	No dedicated page yet	PrepareGenotypes.wdl	Run dosage generation per ancestry/population.
2	Genotype prep	Calculate genotype dosage	No dedicated page yet	calculateGenotypeDosage.wdl	Feed dosages into TensorQTL and SuSiE inputs later.
3	RNA processing	RNA-seq AoU processing (alignment/quant/QC)	RNA-seq AoU Processing	rnaseq_aou.wdl	Branch into eQTL phenotype prep and/or sQTL junction extraction.
4	RNA processing	Aggregate cohort-level RNA outputs (RSEM in WARP; RNA-SeQC2 external for now)	No dedicated page yet	aggregate_rsem_results.wdl	Use aggregated expression/QC summaries for downstream phenotype prep and cohort QC review.
5	eQTL phenotypes	Prepare eQTL phenotype BED + phenotype PCs	No dedicated page yet	prepare_eQTL.wdl	Merge covariates for eQTL TensorQTL run.
6	sQTL phenotypes	Extract junctions from BAM	Leafcutter BAM to Junctions	leafcutter_bam_to_junc.wdl	Cluster junctions to build sQTL phenotype matrices.
7	sQTL phenotypes	Cluster junctions + generate leafcutter outputs	Leafcutter Clustering	leafcutter_cluster.wdl	If needed, run separate phenotype-group generation before covariate merge.
8	sQTL phenotypes	Prepare sQTL phenotype BED + PCs	No dedicated page yet	prepare_sQTL.wdl	Use splicing BED/PC outputs for sQTL covariate merge and TensorQTL.
9	sQTL metadata	Calculate phenotype groups	Calculate Phenotype Groups	CalculatePhenotypeGroups.wdl	Merge covariates for sQTL TensorQTL run.
10	Covariates	Merge covariates (genotype PCs + phenotype PCs ± groups)	No dedicated page yet	MergeCovariates.wdl	Run TensorQTL cis permutations for eQTL/sQTL.
11	Association	TensorQTL cis permutations	No dedicated page yet	tensorqtl_cis_permutations.wdl	Recalculate FDR and prepare significant loci for fine-mapping.
12	Fine-mapping prep	FDR recalculation + SuSiE input preparation, including required AF calculation and genotype dosage checks for downstream aggregation/annotation	No dedicated page yet	calculateAF.wdl	Run SuSiE per phenotype window.
13	Fine-mapping	SuSiE fine-mapping	SuSiE Fine-Mapping Workflow	susieR_workflow.wdl	Aggregate SuSiE outputs across phenotypes.
14	Aggregation	Aggregate SuSiE outputs and annotate	Aggregate SuSiE Workflow	AggregateSusieWorkflow.wdl	Consume required AF outputs from step 12 for interpretation/reporting.

Practical Run Notes

• eQTL path: 0 → 1 → 2 → 3 → 4 → 5 → 10 → 11 → 12 → 13 → 14.
• sQTL path: 0 → 1 → 2 → 3 → 4 → 6 → 7 → 8 → 9 (if required) → 10 → 11 → 12 → 13 → 14.
• Key dependency: susieR_workflow expects TensorQTL-derived significant loci plus dosage inputs from earlier genotype steps.
• Phenotype groups: required for many sQTL TensorQTL configurations; for eQTL they are typically not required.

Additional Processing Notes

• FDR, AF, and SuSiE prep: after TensorQTL, recalculate FDR, filter significant loci (commonly 0.05), calculate AFs, and format SuSiE-ready inputs.
• SuSiE runtime guidance: preemptible VMs can reduce cost; pinned Docker SHAs improve reproducibility.
• Aggregation inputs: for AggregateSusieWorkflow, use fine-mapped SusieParquet outputs and required AF resources from step 12; do not use full/all-tested parquet outputs.
• RNA-level aggregation status: aggregate_rsem_results.wdl is available in WARP for cohort-level RSEM aggregation; an equivalent aggregate rnaseqc2 workflow is not yet available in WARP and currently requires external processing.

Acknowledgements

The original versions of these workflows were either created by the GTEx Consortium (see their GTEx GitHub repository) or by the Dr. Stephen Montgomery Lab at Stanford University. Most of the eQTL scripts originated from the AoU-Multiomics-Analysis repository.

This pipeline builds upon extensive work by the Stephen Montgomery Lab at Stanford University. Special thanks to:

• Evin Padhi
• Jon Nguyen

for developing foundational versions of many scripts and workflows used in this analysis.

Additional integration, optimization, and workflow migration were performed by the All of Us Multiomics and the Broad Pipeline Development teams as part of the WARP workflow suite.

Feedback

Please help us make our tools better by filing an issue in WARP; we welcome pipeline-related suggestions or questions.