Negative log counts values from registration_pseudobulk()

[kinnaryshah]

Thank you for developing this great package! @lahuuki and I discussed this issue and we wanted to get your input here. When I use the registration_pseudobulk() function on my own dataset and on the DLPFC dataset, I get a small percentage of values in the lognormalized pseudobulked matrix as negatives. Almost all of these values are 0 in the counts pseudobulked matrix. Is this behavior expected? I saw that the edgeR::cpm() function uses an average pseudocount of 1, so there are likely some values less than 1.

I bring this up for two reasons: 1) Are these negatives expected? If so, is this okay for running DE? 2) When visualizing the data, having negative logcounts values is hard to interpret. I would prefer to use the logcounts scale for visualization instead of the counts scale. My first solution would be to add a larger prior count, like 3, to prevent negatives while preserving the distribution of the data. Would you recommend this for the DE analysis as well?

spe <- spatialLIBD::fetch_data(type = "spatialDLPFC_Visium")
spe_pseudo <- registration_pseudobulk(spe, var_registration = "BayesSpace_harmony_09", var_sample_id = "sample_id" )
sum(logcounts(spe_pseudo)<0)

Which shows 170,257/3,276,300 values are negative.

[kinnaryshah]

R version 4.3.2 Patched (2024-02-08 r85876)
Platform: x86_64-conda-linux-gnu (64-bit)
Running under: Rocky Linux 9.4 (Blue Onyx)

Matrix products: default
BLAS: /jhpce/shared/community/core/conda_R/4.3.x/R/lib64/R/lib/libRblas.so
LAPACK: /jhpce/shared/community/core/conda_R/4.3.x/R/lib64/R/lib/libRlapack.so; LAPACK version 3.11.0

locale:
[1] LC_CTYPE=en_US.UTF-8 LC_NUMERIC=C
[3] LC_TIME=en_US.UTF-8 LC_COLLATE=en_US.UTF-8
[5] LC_MONETARY=en_US.UTF-8 LC_MESSAGES=en_US.UTF-8
[7] LC_PAPER=en_US.UTF-8 LC_NAME=C
[9] LC_ADDRESS=C LC_TELEPHONE=C
[11] LC_MEASUREMENT=en_US.UTF-8 LC_IDENTIFICATION=C

time zone: US/Eastern
tzcode source: system (glibc)

attached base packages:
[1] stats4 stats graphics grDevices datasets utils methods
[8] base

other attached packages:
[1] dplyr_1.1.4 gridExtra_2.3
[3] spatialLIBD_1.19.6 projectR_1.18.0
[5] synchronicity_1.3.10 bigmemory_4.6.4
[7] cowplot_1.1.3 patchwork_1.2.0
[9] bluster_1.12.0 igraph_2.1.1
[11] RColorBrewer_1.1-3 pheatmap_1.0.12
[13] scran_1.30.2 Matrix_1.6-5
[15] scRNAseq_2.16.0 here_1.0.1
[17] ggspavis_1.9.10 scater_1.30.1
[19] ggplot2_3.5.1 scuttle_1.12.0
[21] SpatialExperiment_1.12.0 SingleCellExperiment_1.24.0
[23] SummarizedExperiment_1.32.0 Biobase_2.62.0
[25] GenomicRanges_1.54.1 GenomeInfoDb_1.38.5
[27] IRanges_2.36.0 S4Vectors_0.40.2
[29] BiocGenerics_0.48.1 MatrixGenerics_1.14.0
[31] matrixStats_1.2.0

loaded via a namespace (and not attached):
[1] later_1.3.2 BiocIO_1.12.0
[3] bitops_1.0-7 filelock_1.0.3
[5] tibble_3.2.1 XML_3.99-0.16.1
[7] lifecycle_1.0.4 edgeR_4.0.14
[9] doParallel_1.0.17 rprojroot_2.0.4
[11] lattice_0.22-5 ensembldb_2.26.0
[13] magrittr_2.0.3 sass_0.4.8
[15] plotly_4.10.4 limma_3.58.1
[17] jquerylib_0.1.4 yaml_2.3.8
[19] metapod_1.10.1 httpuv_1.6.14
[21] NMF_0.28 ggside_0.2.3
[23] askpass_1.2.0 sessioninfo_1.2.2
[25] reticulate_1.35.0 DBI_1.2.1
[27] golem_0.4.1 abind_1.4-5
[29] zlibbioc_1.48.0 purrr_1.0.2
[31] AnnotationFilter_1.26.0 RCurl_1.98-1.14
[33] rappdirs_0.3.3 circlize_0.4.15
[35] GenomeInfoDbData_1.2.11 ggrepel_0.9.5
[37] irlba_2.3.5.1 umap_0.2.10.0
[39] MatrixModels_0.5-3 RSpectra_0.16-1
[41] dqrng_0.3.2 DelayedMatrixStats_1.24.0
[43] codetools_0.2-19 DelayedArray_0.28.0
[45] DT_0.31 xml2_1.3.6
[47] shape_1.4.6 tidyselect_1.2.0
[49] ScaledMatrix_1.10.0 viridis_0.6.5
[51] shinyWidgets_0.8.1 BiocFileCache_2.10.1
[53] GenomicAlignments_1.38.2 jsonlite_1.8.8
[55] GetoptLong_1.0.5 BiocNeighbors_1.20.2
[57] ellipsis_0.3.2 ggalluvial_0.12.5
[59] iterators_1.0.14 foreach_1.5.2
[61] tools_4.3.2 progress_1.2.3
[63] Rcpp_1.0.12 glue_1.7.0
[65] SparseArray_1.2.3 withr_3.0.0
[67] BiocManager_1.30.22 fastmap_1.1.1
[69] fansi_1.0.6 openssl_2.1.1
[71] digest_0.6.34 rsvd_1.0.5
[73] R6_2.5.1 mime_0.12
[75] colorspace_2.1-0 biomaRt_2.58.2
[77] RSQLite_2.3.5 config_0.3.2
[79] tidyr_1.3.1 utf8_1.2.4
[81] generics_0.1.3 data.table_1.15.4
[83] rtracklayer_1.62.0 htmlwidgets_1.6.4
[85] prettyunits_1.2.0 httr_1.4.7
[87] S4Arrays_1.2.0 pkgconfig_2.0.3
[89] gtable_0.3.4 tsne_0.1-3.1
[91] registry_0.5-1 blob_1.2.4
[93] ComplexHeatmap_2.18.0 XVector_0.42.0
[95] htmltools_0.5.7 clue_0.3-65
[97] ProtGenerics_1.34.0 scales_1.3.0
[99] attempt_0.3.1 png_0.1-8
[101] bigmemory.sri_0.1.8 uuid_1.2-0
[103] reshape2_1.4.4 rjson_0.2.21
[105] curl_5.2.3 GlobalOptions_0.1.2
[107] cachem_1.0.8 stringr_1.5.1
[109] BiocVersion_3.18.1 parallel_4.3.2
[111] vipor_0.4.7 AnnotationDbi_1.64.1
[113] restfulr_0.0.15 pillar_1.9.0
[115] grid_4.3.2 vctrs_0.6.5
[117] promises_1.2.1 BiocSingular_1.18.0
[119] dbplyr_2.4.0 beachmat_2.18.0
[121] xtable_1.8-4 cluster_2.1.6
[123] paletteer_1.6.0 beeswarm_0.4.0
[125] GenomicFeatures_1.54.3 magick_2.8.2
[127] cli_3.6.2 locfit_1.5-9.8
[129] compiler_4.3.2 Rsamtools_2.18.0
[131] rngtools_1.5.2 rlang_1.1.3
[133] crayon_1.5.2 rematch2_2.1.2
[135] plyr_1.8.9 ggbeeswarm_0.7.2
[137] stringi_1.8.3 gridBase_0.4-7
[139] viridisLite_0.4.2 BiocParallel_1.36.0
[141] munsell_0.5.0 Biostrings_2.70.2
[143] lazyeval_0.2.2 benchmarkme_1.0.8
[145] ExperimentHub_2.10.0 hms_1.1.3
[147] sparseMatrixStats_1.14.0 bit64_4.0.5
[149] KEGGREST_1.42.0 statmod_1.5.0
[151] shiny_1.8.0 interactiveDisplayBase_1.40.0
[153] AnnotationHub_3.10.0 ROCR_1.0-11
[155] memoise_2.0.1 bslib_0.6.1
[157] benchmarkmeData_1.0.4 bit_4.0.5