Command-Line Options

Here is the list of all fastENLOC command-line options

1. Required

• -total_variant | -tv number_of_total_variants: required for all running options, except when $p_1$, $p_2$, and $p_{12}$ are specified.

1.1 Probabilistic Fine-mapping Input

• -eqtl | -e eqtl_probabilitic_annotation
• -gwas | -g gwas_probabilistic_annotation

1.2 Summary Statistics Input

• -summary | -sum combined_summary_input: combined summary statistics input
• -gs gwas_summary_statistics: GWAS summary statistics that used in the hybrid option

1.3 Legacy GWAS probabilistic input

• -go gwas_probabilistic_legacy_annotation: for GWAS summary statistics processed by TORUS

1.4 Tissue Specification

• -tissue | -t tissue: required only when multi-tissue QTL annotations are used

2. Strongly Recommended

• -prefix output_prefix: specify the output file prefix and directory. Use this option to avoid overwriting the result files with default names.

3. Optional

3.1 Bypassing Enrichment Analysis

• -p1 p1_value: frequency of GWAS-only SNPs
• -p2 p2_value: frequency of QTL-only SNPs

• -p12 p12_value: frequency of colocalized SNPs

• -coloc_default_prior: a handy option to specify
  • $p_1 = 1\times 10^{-4}$
  • $p_2 = 1\times 10^{-4}$
  • $p_{12} = 1\times 10^{-5}$

• -a0 a0_value
• -a1 a1_value

3.2 Bypassing Colocalization Computation (Enrichment Analysis Only)

• --enrich_only | --enrich: performing enrichment analysis only without further computing colcoalization probabilities

3.3 Multi-thread Processing

• -thread number_of_parallel_threads: for performance improvement. But for most application cases, single-thread processing is quite efficient

3.4 Colocalization Algorithm Options

• --approx | --legacy: use legacy computational algorithm implemented in the original ENLOC described in Wen et al., 2017.
• --exact: the default computational algorithm for computing colocalization probabilities, implemented since version 2.

• -shrinkage | -s shrinkage_coefficient: shrinkage coefficient for enrichment analysis, the default shrinkage coefficient is 1, corresponding to a ${\rm N}(0,1)$ prior on $\alpha_1$. Larger coefficient value results in stronger shrinkage. Set -s 0 for no shrinkage on $\alpha_1$.

• -impute | -imp multiple_imputation_rounds: specify the number imputations for estimating the enrichment parameters. The default value is 25, which is generally sufficient.

3.5 Output Options

• --output_all | --all: output colocalization probabilities for all SNPs. By default, the SNP output contains only the subset with SCP (i.e., SNP-level colocalization probability) $ \ge 10^{-4}$.

3.6 Miscellaneous Options

• -cap_a1 a1_upper_limit: overriding the estimated $\hat \alpha_1$ value with a pre-defined upper limit. It is mostly used for sensitivity analysis to examine how $\hat \alpha_1$ affects colocalization computation.
• -seed random_seed: user defined random seed for multi-imputation in enrichment analysis.
• -sdy_e: specify a single grid for BF computation in eQTL data
• -sdy_g: specify a single grid for BF computation in GWAS data
• -conv: convert summary statistic input to probabilistic fine-mapping format without performing any computational tasks (enrichment analysis or colocalization computation).