Import precomputed graph

[Dannyxu123]

Hi! it's a great tool to use. However, After adding precomputated graph from an integrated Seurat Object (integrated using RPCA) into milo Object, makeNhoods could not identify valid graph for processing. However, when checking the milo Object, the graph is there. I am not sure where to begin troubleshooting.

## your code
test<-as.SingleCellExperiment(Merge.combined,assay = "integrated")
test_milo <- Milo(test)
graph(test_milo) <- buildFromAdjacency(Merge.combined@graphs$integrated_snn,k=20)
test_milo <- makeNhoods(test_milo, prop = 0.1, k = 30, d=30, refined = TRUE, reduced_dims = "UMAP")

**Errors
Checking valid object
Error in makeNhoods(test_milo, prop = 0.1, k = 30, d = 30, refined = TRUE, :
Not a valid Milo object - graph is missing. Please run buildGraph() first.

Session info

sessionInfo()
R version 4.2.2 (2022-10-31)
Platform: x86_64-pc-linux-gnu (64-bit)
Running under: Ubuntu 22.04.2 LTS

Matrix products: default
BLAS: /usr/lib/x86_64-linux-gnu/openblas-pthread/libblas.so.3
LAPACK: /usr/lib/x86_64-linux-gnu/openblas-pthread/libopenblasp-r0.3.20.so

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

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

other attached packages:
[1] miloR_1.6.0 edgeR_3.40.2 limma_3.54.2
[4] patchwork_1.2.0 pheatmap_1.0.12 ggsci_3.0.0
[7] dplyr_1.1.4 cowplot_1.1.3 ggalluvial_0.12.5
[10] biomaRt_2.54.1 ddqcR_0.1.0 scCustomize_1.1.3
[13] ggrepel_0.9.5 uwot_0.1.16 hexbin_1.28.3
[16] circlize_0.4.16 ComplexHeatmap_2.14.0 nabor_0.5.0
[19] SeuratObject_4.1.3 Seurat_4.3.0 rhdf5_2.42.1
[22] SummarizedExperiment_1.28.0 Biobase_2.58.0 MatrixGenerics_1.10.0
[25] Rcpp_1.0.12 Matrix_1.6-5 GenomicRanges_1.50.2
[28] GenomeInfoDb_1.34.9 IRanges_2.32.0 S4Vectors_0.36.2
[31] BiocGenerics_0.44.0 matrixStats_1.2.0 data.table_1.15.0
[34] stringr_1.5.1 plyr_1.8.9 magrittr_2.0.3
[37] ggplot2_3.4.4 gtable_0.3.4 gtools_3.9.5
[40] gridExtra_2.3 ArchR_1.0.2

loaded via a namespace (and not attached):
[1] utf8_1.2.4 spatstat.explore_3.2-6 reticulate_1.35.0
[4] tidyselect_1.2.0 RSQLite_2.3.5 AnnotationDbi_1.60.2
[7] htmlwidgets_1.6.4 BiocParallel_1.32.6 Rtsne_0.17
[10] ScaledMatrix_1.6.0 munsell_0.5.0 codetools_0.2-18
[13] ica_1.0-3 future_1.33.1 miniUI_0.1.1.1
[16] withr_3.0.0 spatstat.random_3.2-2 colorspace_2.1-0
[19] progressr_0.14.0 filelock_1.0.3 rstudioapi_0.15.0
[22] SingleCellExperiment_1.20.1 ROCR_1.0-11 tensor_1.5
[25] listenv_0.9.1 labeling_0.4.3 GenomeInfoDbData_1.2.9
[28] polyclip_1.10-6 farver_2.1.1 bit64_4.0.5
[31] parallelly_1.36.0 vctrs_0.6.5 generics_0.1.3
[34] BiocFileCache_2.6.1 timechange_0.3.0 R6_2.5.1
[37] doParallel_1.0.17 graphlayouts_1.1.0 ggbeeswarm_0.7.2
[40] clue_0.3-65 rsvd_1.0.5 locfit_1.5-9.8
[43] bitops_1.0-7 rhdf5filters_1.10.1 spatstat.utils_3.0-4
[46] cachem_1.0.8 DelayedArray_0.24.0 promises_1.2.1
[49] scales_1.3.0 ggraph_2.1.0 beeswarm_0.4.0
[52] beachmat_2.14.2 Cairo_1.6-2 globals_0.16.2
[55] goftest_1.2-3 spam_2.10-0 tidygraph_1.3.1
[58] rlang_1.1.3 GlobalOptions_0.1.2 splines_4.2.2
[61] lazyeval_0.2.2 spatstat.geom_3.2-8 reshape2_1.4.4
[64] abind_1.4-5 httpuv_1.6.14 tools_4.2.2
[67] ellipsis_0.3.2 RColorBrewer_1.1-3 ggridges_0.5.6
[70] progress_1.2.3 zlibbioc_1.44.0 purrr_1.0.2
[73] RCurl_1.98-1.14 prettyunits_1.2.0 deldir_2.0-2
[76] viridis_0.6.5 pbapply_1.7-2 GetoptLong_1.0.5
[79] zoo_1.8-12 cluster_2.1.4 scattermore_1.2
[82] lmtest_0.9-40 RANN_2.6.1 fitdistrplus_1.1-11
[85] hms_1.1.3 mime_0.12 xtable_1.8-4
[88] XML_3.99-0.16.1 shape_1.4.6 compiler_4.2.2
[91] tibble_3.2.1 KernSmooth_2.23-20 crayon_1.5.2
[94] htmltools_0.5.7 ggfun_0.1.4 later_1.3.2
[97] ggprism_1.0.4 tidyr_1.3.1 lubridate_1.9.3
[100] DBI_1.2.2 tweenr_2.0.2 dbplyr_2.4.0
[103] rappdirs_0.3.3 MASS_7.3-58.1 cli_3.6.2
[106] dotCall64_1.1-1 igraph_2.0.2 forcats_1.0.0
[109] pkgconfig_2.0.3 tidydr_0.0.5 sp_2.1-3
[112] plotly_4.10.4 spatstat.sparse_3.0-3 xml2_1.3.6
[115] paletteer_1.6.0 foreach_1.5.2 vipor_0.4.7
[118] XVector_0.38.0 snakecase_0.11.1 digest_0.6.34
[121] sctransform_0.4.1 RcppAnnoy_0.0.22 janitor_2.2.0
[124] spatstat.data_3.0-4 Biostrings_2.66.0 leiden_0.4.3.1
[127] curl_5.2.0 shiny_1.8.0 rjson_0.2.21
[130] lifecycle_1.0.4 nlme_3.1-160 jsonlite_1.8.8
[133] Rhdf5lib_1.20.0 BiocNeighbors_1.16.0 viridisLite_0.4.2
[136] fansi_1.0.6 pillar_1.9.0 lattice_0.20-45
[139] ggrastr_1.0.2 KEGGREST_1.38.0 fastmap_1.1.1
[142] httr_1.4.7 survival_3.4-0 glue_1.7.0
[145] png_0.1-8 iterators_1.0.14 bit_4.0.5
[148] ggforce_0.4.2 stringi_1.8.3 rematch2_2.1.2
[151] blob_1.2.4 BiocSingular_1.14.0 memoise_2.0.1
[154] irlba_2.3.5.1 future.apply_1.11.1

[MikeDMorgan]

Do you get the same error if you explicitly use miloR::graph()?

[Dannyxu123]

Do you get the same error if you explicitly use miloR::graph()?

Yes, I get the same error. And here's the info I get when I run buildFromAdjacency:
Adding nhoodDistances to Milo object
Warning message:
In .M2v(x) : sparse->dense coercion: allocating vector of size 6.3 GiB

I am not sure if this indicates graph is added successfully or not?

[MikeDMorgan]

OK - your version of Milo is very out of date (1.6.0 vs. current 2.0.0).

Secondly, the error is an out of memory error, so no, the graph isn't being added. Could you paste the code with the traceback immediately following it?

[mdpatric]

Hello, I believe the issue is after running buildFromAdjacency(seurat.combined@graphs$SCT_snn,k=20) it adds another milo object into the graph slot with it's own graph slot with an iGraph object. After running milo@graph[["graph"]] <- milo@graph[["graph"]]@graph[["graph"]] mine runs.

This issue is also addressed by simply adding
miloR::graph(test_milo) <- miloR::graph(buildFromAdjacency(Merge.combined@graphs$integrated_snn,k=20))

[Dannyxu123]

Hello, I believe the issue is after running buildFromAdjacency(seurat.combined@graphs$SCT_snn,k=20) it adds another milo object into the graph slot with it's own graph slot with an iGraph object. After running milo@graph[["graph"]] <- milo@graph[["graph"]]@graph[["graph"]] mine runs.

This issue is also addressed by simply adding miloR::graph(test_milo) <- miloR::graph(buildFromAdjacency(Merge.combined@graphs$integrated_snn,k=20))

This works! thank you a lot! However, makeNhoods seems to lead to another problem. I guess I will update the milo first
##Code
test_milo <- makeNhoods(test_milo, prop = 0.1, k =20, d=30, refined = TRUE, reduced_dims = "UMAP")
##Info
Checking valid object
Running refined sampling with reduced_dim
Error in .subset_to_index(subset, precomputed, byrow = TRUE) :
'subset' indices out of range of 'x'

[MikeDMorgan]

@Dannyxu123 - do not use UMAP for constructing your graph - the distances aren't valid. Use an alternative reduced dimensional space where the distances are retained, e.g. PCA or diffusion map.

[MikeDMorgan]

Hello, I believe the issue is after running buildFromAdjacency(seurat.combined@graphs$SCT_snn,k=20) it adds another milo object into the graph slot with it's own graph slot with an iGraph object. After running milo@graph[["graph"]] <- milo@graph[["graph"]]@graph[["graph"]] mine runs.

This issue is also addressed by simply adding miloR::graph(test_milo) <- miloR::graph(buildFromAdjacency(Merge.combined@graphs$integrated_snn,k=20))

In general, it is not good practise to access objects using the @ accessor - if these slot names change then your code will break. However, using the appropriate getter and setter methods ensures that functionality is retained regardless of what the slot names are.