Stabilises the fitted per-(gene, focal, neighbour) proximity slopes with
multivariate adaptive shrinkage, populating neighbourSlopes().
Arguments
- object
A PACEFit from
paceModel().- ...
Further arguments passed to the shrinkage step, notably
null_correlation(defaultTRUE: estimate the correlation between focal cell types under the null and pass it to mash asV;FALSEtreats them as independent),data_driven, andshrink_threads.shrink_threads(default 1) is how many R processes shrink the neighbour slices at once. The slices are independent and carry about 90% of the work, so this is where the time goes: on the full breast cancer cohort it takes the shrinkage from 55.4 to 21.1 seconds, and the called slopes agree with the serial path to 8.3e-13 with identical calls and no sign flips.It is not, however, the serial computation. With
data_driven = TRUEthe parallel path cannot reproduce the serial one:cov_pca()draws its starting vectors from the RNG, and a serial run consumes that stream slice by slice. Above 1 each slice is seeded with its own index, so the parallel result is reproducible run to run but differs from the serial one – by ~1e-12 at cohort scale, but by up to ~1e-6 on small slices, where there is less data to swamp the difference. Serial is the default because it is the stream the package's fixtures were made with, and because the parallel path raised a BiocParallel reducer error on one supported R version that could not be reproduced on any other. Raise it when a large cohort's shrinkage is the bottleneck. Withdata_driven = FALSEnothing draws from the stream and the two are identical.
Details
One mash model is fitted per neighbour cell type, and every one of them is fitted over the same genes. mash calibrates against the genes it is given, so a common gene set keeps the lfsr comparable across neighbours; a gene that cannot be used for one neighbour is dropped for all of them, with a message.
Examples
fit <- readRDS(system.file("extdata", "pace_fit_example.rds", package = "PACE"))
fit <- paceShrink(fit)
#> - Computing 278 x 313 likelihood matrix.
#> - Likelihood calculations took 0.09 seconds.
#> - Fitting model with 313 mixture components.
#> - Model fitting took 0.22 seconds.
#> - Computing posterior matrices.
#> - Computation allocated took 0.00 seconds.
#> [mashr] B_Cell: 278 genes shrunk; gene-focal calls (lfsr<0.05) = 0, expected false sign = 0.0
#> [mashr] term 'Dendritic_Cell': fewer than two varying conditions, using independence
#> - Computing 278 x 92 likelihood matrix.
#> - Likelihood calculations took 0.02 seconds.
#> - Fitting model with 92 mixture components.
#> - Model fitting took 0.06 seconds.
#> - Computing posterior matrices.
#> - Computation allocated took 0.00 seconds.
#> [mashr] Dendritic_Cell: 278 genes shrunk; gene-focal calls (lfsr<0.05) = 0, expected false sign = 0.0
#> - Computing 278 x 404 likelihood matrix.
#> - Likelihood calculations took 0.11 seconds.
#> - Fitting model with 404 mixture components.
#> - Model fitting took 0.29 seconds.
#> - Computing posterior matrices.
#> - Computation allocated took 0.00 seconds.
#> [mashr] Endothelial: 278 genes shrunk; gene-focal calls (lfsr<0.05) = 2, expected false sign = 0.0
#> - Computing 278 x 417 likelihood matrix.
#> - Likelihood calculations took 0.12 seconds.
#> - Fitting model with 417 mixture components.
#> - Model fitting took 0.47 seconds.
#> - Computing posterior matrices.
#> - Computation allocated took 0.00 seconds.
#> [mashr] Macrophage: 278 genes shrunk; gene-focal calls (lfsr<0.05) = 5, expected false sign = 0.0
#> [mashr] term 'Myoepithelial': fewer than two varying conditions, using independence
#> - Computing 278 x 92 likelihood matrix.
#> - Likelihood calculations took 0.00 seconds.
#> - Fitting model with 92 mixture components.
#> - Model fitting took 0.05 seconds.
#> - Computing posterior matrices.
#> - Computation allocated took 0.00 seconds.
#> [mashr] Myoepithelial: 278 genes shrunk; gene-focal calls (lfsr<0.05) = 0, expected false sign = 0.0
#> - Computing 278 x 391 likelihood matrix.
#> - Likelihood calculations took 0.11 seconds.
#> - Fitting model with 391 mixture components.
#> - Model fitting took 0.25 seconds.
#> - Computing posterior matrices.
#> - Computation allocated took 0.00 seconds.
#> [mashr] Stromal: 278 genes shrunk; gene-focal calls (lfsr<0.05) = 3, expected false sign = 0.0
#> - Computing 278 x 430 likelihood matrix.
#> - Likelihood calculations took 0.12 seconds.
#> - Fitting model with 430 mixture components.
#> - Model fitting took 0.26 seconds.
#> - Computing posterior matrices.
#> - Computation allocated took 0.00 seconds.
#> - Computing 278 x 628 likelihood matrix.
#> - Likelihood calculations took 0.17 seconds.
#> - Fitting model with 628 mixture components.
#> - Model fitting took 0.36 seconds.
#> - Computing posterior matrices.
#> - Computation allocated took 0.00 seconds.
#> [mashr] T_Cell: 278 genes shrunk; gene-focal calls (lfsr<0.05) = 21, expected false sign = 0.1
#> - Computing 278 x 404 likelihood matrix.
#> - Likelihood calculations took 0.11 seconds.
#> - Fitting model with 404 mixture components.
#> - Model fitting took 0.60 seconds.
#> - Computing posterior matrices.
#> - Computation allocated took 0.01 seconds.
#> - Computing 278 x 590 likelihood matrix.
#> - Likelihood calculations took 0.16 seconds.
#> - Fitting model with 590 mixture components.
#> - Model fitting took 0.86 seconds.
#> - Computing posterior matrices.
#> - Computation allocated took 0.01 seconds.
#> [mashr] Tumour: 278 genes shrunk; gene-focal calls (lfsr<0.05) = 30, expected false sign = 0.1
head(neighbourSlopes(fit))
#> gene focal neighbour term estimate std.error estimate_shrunk
#> 1 ABCC11 B_Cell B_Cell B_Cell -1.968907e-05 0.001376136 0
#> 2 ACTA2 B_Cell B_Cell B_Cell -9.976965e-05 0.004245639 0
#> 3 ACTG2 B_Cell B_Cell B_Cell 1.418293e-03 0.005170493 0
#> 4 ADAM9 B_Cell B_Cell B_Cell 3.277669e-05 0.003072945 0
#> 5 ADGRE5 B_Cell B_Cell B_Cell 8.125370e-04 0.005304314 0
#> 6 ADH1B B_Cell B_Cell B_Cell -3.142864e-05 0.002131910 0
#> sd_shrunk lfsr
#> 1 0 1
#> 2 0 1
#> 3 0 1
#> 4 0 1
#> 5 0 1
#> 6 0 1