Error in geometry::delaunayn(p = spatial_locations[, c(sdimx, sdimy, sdimz), : Received error code 1 from qhull.

[samuelikim]

Hi I'm running into this error after running the default pipeline, here's the full error message:

Error in geometry::delaunayn(p = spatial_locations[, c(sdimx, sdimy, sdimz), :
Received error code 1 from qhull. Qhull error:
QH6239 Qhull precision error: initial Delaunay input sites are cocircular or cospherical. Use option 'Qz' for the Delaunay triangulation or Voronoi diagram of cocircular/cospherical points; it adds a point "at infinity". Alternatively use option 'QJ' to joggle the input. Use option 'Qs' to search all points for the initial simplex.

input sites with last coordinate projected to a paraboloid

• p4(v5): 2.1e+03 -3.4e+03 -0.8 5e-11
• p1(v4): 2.1e+03 -3.4e+03 0.8 2.7e+03
• p0(v3): 2.1e+03 -3.4e+03 0 3.4e+03
• p2(v2): 2.1e+03 -3.4e+03 -0.8 2.1e+02
• p3(v1): 2.1e+03 -3.4e+03 -0.8 5e-11

While executing: | qhull d Qbb T0 Pp Qt
Options selected for Qhull 2020.2.r 2020/08/31:
run-id 1295723442 delaunay Qbbound-last Pprecision-ignore Qtriangulate
_pre-merge _zero-centrum Pgood _max-width 2.8 Error-roundoff 6.9e-12
_one-merge 6.2e-11 _near-inside 3.1e-10 Visible-distance 4.2e-11
U-max-coplanar 4.2e-11 Width-ou

My code isa s follows

run_fastReseg_pipeline <- function(
base_dir,
expr_file_name,
tx_file_name,
output_dir_name = "res2_fullPipeline",
pixel_size = 0.12028,
zstep_size = 0.8,
n_cores = 7,
fov_range = NULL # NEW ARGUMENT
) {

2. Set file paths

expr_file <- file.path(base_dir, expr_file_name)
tx_file <- file.path(base_dir, tx_file_name)

3. Load expression matrix

counts <- fread(expr_file)

--- FILTER BY FOV IF SPECIFIED ---

if (!is.null(fov_range)) {
counts <- counts[fov %in% fov_range, ]
}

Negative probes

neg_cols <- grep("Negative", colnames(counts), value = TRUE)
neg_cols_no_add <- grep("NegativeAdd", neg_cols, value = TRUE, invert = TRUE)

Based on the expression matrix, create a Seurat object, run SCT, up to Clusters

cell_ids <- paste0('c_1_', counts[['fov']], '_', counts[['cell_ID']])
rownames(counts) <- cell_ids
all_rnas <- grep("fov|cell_ID|Negative|SystemControl|huCART19-IL18-huCD19|IDO2",
colnames(counts), value = TRUE, invert = TRUE)
counts_mat <- as.matrix(counts[, .SD, .SDcols = all_rnas])
rownames(counts_mat) <- cell_ids
sparse_counts <- as(counts_mat, "dgCMatrix")
sparse_counts <- t(sparse_counts) # Transpose after making sparse
seu <- CreateSeuratObject(counts = sparse_counts)

seu <- SCTransform(seu, verbose = FALSE)
seu <- RunPCA(seu, verbose = FALSE, npcs = 50)
seu <- FindNeighbors(seu, reduction = "pca", dims = 1:30)
seu <- FindClusters(seu, resolution = 1.2, algorithm = 4)

4. Load and process transcript file

fullTx <- data.table::fread(tx_file)
if (!is.null(fov_range)) {
fullTx <- fullTx[fov %in% fov_range, ]
}
fullTx[['transcript_id']] <- seq_len(nrow(fullTx))
fullTx <- fullTx[cell_ID != 0, ]
fullTx <- fullTx[, .SD, .SDcols = c('transcript_id', 'cell', 'x_global_px',
'y_global_px', 'z', 'target', 'fov')]

5. Set parameters

clust <- as.character(seu$seurat_clusters)
names(clust) <- colnames(seu)

outDir_full <- file.path(base_dir, output_dir_name)
options(mc.cores = n_cores)

6. Run FastReseg pipeline

refineAll_res <- FastReseg::fastReseg_full_pipeline(
counts = counts_mat,
clust = clust,
refProfiles = NULL,
transcript_df = fullTx,
transDF_fileInfo = NULL,
filepath_coln = 'file_path',
prefix_colns = NULL,
fovOffset_colns = c('stage_Y','stage_X'),
pixel_size = pixel_size,
zstep_size = zstep_size,
transID_coln = 'transcript_id',
transGene_coln = "target",
cellID_coln = "cell",
spatLocs_colns = c('x_global_px', 'y_global_px', 'z'),
extracellular_cellID = NULL,
flagModel_TransNum_cutoff = 50,
molecular_distance_cutoff = 2.7,
cellular_distance_cutoff = NULL,
score_baseline = NULL,
lowerCutoff_transNum = NULL,
higherCutoff_transNum = NULL,
imputeFlag_missingCTs = TRUE,
flagCell_lrtest_cutoff = 5,
svmClass_score_cutoff = -2,
groupTranscripts_method = "dbscan",
spatialMergeCheck_method = "leidenCut",
cutoff_spatialMerge = 0.5,
path_to_output = outDir_full,
save_intermediates = TRUE,
return_perCellData = TRUE,
combine_extra = FALSE,
ctrl_genes = neg_cols_no_add
)

return(refineAll_res)
}

refineAll_res <- run_fastReseg_pipeline(
base_dir = base_dir,
expr_file_name = "test.csv.gz",
tx_file_name = "test.csv.gz",
output_dir_name = "res2_fullPipeline_06202025",
pixel_size = 0.12028,
zstep_size = 0.8,
n_cores = 7
)

[github-actions]

Thank you for contacting us about our tools! To receive faster assistance, kindly email support.spatial@bruker.com with detailed information about your issue. If applicable, attach a screenshot of any encountered errors and include a copy of the modified script in Notepad. Our customer support team will help facilitate a review and resolution of the issue.

Thank you for choosing Bruker Spatial Biology,
Bruker Spatial Biology Dev Team

[bethk-h]

Hi - I'm also running into the same issue.

Has a solution been suggested?

Thanks
Beth

[lidanwu]

Hello @samuelikim and @bethk-h, sorry for the delay in response.

The error message seems to indicate all transcripts under analysis are in same 2D plane and thus error during creation on 3D spatial network.

On one hand, the package does have an automatic switch between 2D vs. 3D network creation to deal with the case when all transcripts of interest are within same z plane and you should some log message like following indicating the switch from 3D to 2D in network analysis.

But the error message you have indicates the input transcript coordinates have exactly the same XY coordinates at (2.1e+03 , -3.4e+03) with difference in z coordinates only -0.8 vs 0 vs 0.8. These means all those transcripts of interest are in 1 line and thus failed during spatial network analysis.

I don't have a immediate solution to this unfortunately. Are you working with thin tissue section (<1 layer of average cell diameter)? If yes, it might not be a bad idea to use fastReseg_flag_all_errors() to identify all problematic transcripts and then trim/remove them from you data directly instead of going through the full refinement pipeline.

[bethk-h]

Thanks @lidanwu for the response.

I've got it to run by removing the z-plane altogether so it only runs 2D network analysis. Does this seem like a reasonable solution for this dataset?

[lidanwu]

@bethk-h , not a bad idea to try. When providing 2D coordinates for the entire pipeline, all the spatial evaluation would be done in 2D including the flagging of both cells and transcripts. This could result in slighting different flagging and may either ignore or pick up slightly more transcripts in the neighborhood of previously problematic transcripts.

[lidanwu]

A fallback solution on spatial network for colinear points is added in PR #57 for package version v1.1.0 and above.

[StevenWijnen]

I am working with FastResed 1.1.1 but still encounter this issue when calling the FastReseg::fastReseg_full_pipeline function.

R version 4.5.2 (2025-10-31)
Platform: x86_64-conda-linux-gnu
Running under: Ubuntu 24.04.3 LTS

Matrix products: default

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

other attached packages:
[1] FastReseg_1.1.1

loaded via a namespace (and not attached):
[1] Matrix_1.7-4 gtable_0.3.6 jsonlite_2.0.0
[4] dplyr_1.1.4 compiler_4.5.2 gtools_3.9.5
[7] spatstat.utils_3.2-0 tidyselect_1.2.1 Rcpp_1.1.0
[10] GiottoClass_0.4.10 scales_1.4.0 lattice_0.22-7
[13] deldir_2.0-4 ggplot2_4.0.1 R6_2.6.1
[16] spatstat.univar_3.1-5 dbscan_1.2.3 generics_0.1.4
[19] igraph_2.2.1 spatstat.geom_3.6-1 backports_1.5.0
[22] polyclip_1.10-7 checkmate_2.3.3 tibble_3.3.0
[25] pillar_1.11.1 RColorBrewer_1.1-3 GiottoUtils_0.2.5
[28] rlang_1.1.6 terra_1.8-86 fs_1.6.6
[31] S7_0.2.1 cli_3.6.5 magrittr_2.0.4
[34] grid_4.5.2 concaveman_1.2.0 lifecycle_1.0.4
[37] vctrs_0.6.5 glue_1.8.0 data.table_1.17.8
[40] farver_2.1.2 spatstat.data_3.1-9 pkgconfig_2.0.3

[lidanwu]

@StevenWijnen Thanks for reporting. If you can share the minimal input data (input arguments and transcript file for the problem FOV) for reproducing the error to our support team, I can take a closer look and see how to resolve this issue. 

[lidanwu]

@StevenWijnen I now have a better fix in PR #62 for this collinearity issue. Would you mind to try it out on your data to see if it resolves your problem? You can install the code with the code below.

devtools::install_github("Nanostring-Biostats/FastReseg", 
                         build_vignettes = TRUE, ref = "iss45_delaunay")