Simulated multi-outcome QTL example — multiomic

Simulated multi-outcome QTL data for mfsusie() covering DNA methylation (functional, T = 64), RNA-seq (functional, T = 32), and two scalar QTLs (eQTL, pQTL). Two causal SNPs shared across all four outcomes; per-outcome shapes and signs differ.

Usage

multiomic_example

Format

A list with components

X: n x p genotype matrix (p = 150) sliced from susieR::N3finemapping$X.
Y_list: named list of four outcomes: dnam (n x 64), rna (n x 32), eqtl (n x 1), pqtl (n x 1).
pos_list: named list of position vectors (CpG bp, exon-body indices, scalar dummy 1L for the QTLs).
causal_snps: integer vector of shared causal SNPs.
description: free-text description.

Source

Simulated. See data-raw/make_data.R.

Examples

# \donttest{
data(multiomic_example)
fit <- mfsusie(multiomic_example$X, multiomic_example$Y_list,
               pos = multiomic_example$pos_list, L = 15, L_greedy = 5,
               verbose = TRUE)
#> HINT: ncol(Y) is not 2^J or positions are unevenly spaced; interpolated to a regular dyadic grid.
#> iter          ELBO       delta   sigma2      mem      V  extras
#>    1   -78950.9731           -   [0.998, 0.999, 1.000]   0.19 GB  [1.00e+00, 1.00e+00, 1.00e+00, 1.00e+00, 1.00e+00]  pi_null=[1.00, 1.00]
#> iter   2: max|dPIP|=1.95e-02, V=[1.00e+00, 1.00e+00, 1.00e+00, 1.00e+00, 1.00e+00] [mem: 0.19 GB]
#> iter   3: max|dPIP|=6.97e-03, V=[1.00e+00, 1.00e+00, 1.00e+00, 1.00e+00, 1.00e+00] [mem: 0.19 GB]
#> iter   4: max|dPIP|=6.53e-03, V=[1.00e+00, 1.00e+00, 1.00e+00, 1.00e+00, 1.00e+00] [mem: 0.19 GB]
#> iter   5: max|dPIP|=7.53e-03, V=[1.00e+00, 1.00e+00, 1.00e+00, 1.00e+00, 1.00e+00] [mem: 0.19 GB]
#> iter   6: max|dPIP|=1.10e-02, V=[1.00e+00, 1.00e+00, 1.00e+00, 1.00e+00, 1.00e+00] [mem: 0.19 GB]
#> iter   7: max|dPIP|=1.72e-02, V=[1.00e+00, 1.00e+00, 1.00e+00, 1.00e+00, 1.00e+00] [mem: 0.19 GB]
#> iter   8: max|dPIP|=1.89e-02, V=[1.00e+00, 1.00e+00, 1.00e+00, 1.00e+00, 1.00e+00] [mem: 0.19 GB]
#> iter   9: max|dPIP|=2.90e-02, V=[1.00e+00, 1.00e+00, 1.00e+00, 1.00e+00, 1.00e+00] -- converged (stalled) [mem: 0.19 GB]
#> WARNING: PIP convergence stalled (no improvement in 5 iterations); returning current state.
#> [L_greedy] 1 round, greedy_lbf_cutoff=0.100, final L=5
#> round  L     min(lbf)   action
#> 1      5     0.000      saturated
fit_s <- mf_post_smooth(fit, method = "TI")
#> HINT: method = 'TI' is a wavelet smoother and adds no power for outcome 3 (T_m = 1, scalar). Falling back to method = 'scalewise' for that outcome.
#> HINT: method = 'TI' is a wavelet smoother and adds no power for outcome 4 (T_m = 1, scalar). Falling back to method = 'scalewise' for that outcome.
mfsusie_plot(fit_s)

# }