Error estimating prior segmentation polygons

[GMaciag]

Hi Viktor

This looks like such a great package, thank you very much for it. I tried using it on my own data while supplying prior segmentation from a DAPI staining, but I get an error. I tried both segmenting on my own with Mesmer (DeepCell) as well as in ImageJ following your tutorial in Readme. Both times it failed with the same error message.

Command:
baysor run -x x_c -y y_c -z z_c --gene gene -o ./results_C1-2/ -i 500 --force-2d -p --n-clusters 7 ../C1-2_results.txt ../C1-2_segmented_fiji.tif

Output:

┌ Warning: No config file provided. Using default parameters.
└ @ Baysor ~/.julia/packages/Baysor/tda03/src/cli/main.jl:96
[14:22:30] Info: (2021-12-07) Run Baysor v0.5.1
[14:22:30] Info: Loading data...
[14:22:48] Info: Loaded 126550 transcripts
[14:22:49] Info: Loading segmentation mask...
[14:23:05] Info: Done
[14:23:05] Info: Estimating prior segmentation polygons...
[14:23:10] Error: MethodError(LightGraphs.kruskal_mst, ({165106, 163115} undirected simple Int64 graph with Float64 weights,), 0x0000000000007b62)
|
|                 border_mst(border_points::Vector{Vector{UInt32}}; min_edge_length::Float64, max_edge_length::Float64) at boundary_estimation.jl:65
|                 border_mst(border_points::Vector{Vector{UInt32}}) at boundary_estimation.jl:42
|                 _broadcast_getindex_evalf at broadcast.jl:670 [inlined]
|                 _broadcast_getindex at broadcast.jl:643 [inlined]
|                 getindex at broadcast.jl:597 [inlined]
|                 copy at broadcast.jl:943 [inlined]
|                 materialize at broadcast.jl:904 [inlined]
|                 extract_polygons_from_label_grid(grid_labels::Matrix{UInt32}; min_border_length::Int64, shape_method::Symbol, max_dev::Float64, exclude_labels::Vector{Int64}, offset::Vector{Float64}, grid_step::Float64) at boundary_estimation.jl:231
|                 (::Baysor.var"#extract_polygons_from_label_grid##kw")(::NamedTuple{(:grid_step,), Tuple{Float64}}, ::typeof(Baysor.extract_polygons_from_label_grid), grid_labels::Matrix{UInt32}) at boundary_estimation.jl:223
|                 load_prior_segmentation(df_spatial::DataFrames.DataFrame, args::Dict{String, Any}) at main.jl:297
|                 load_and_preprocess_data!(args::Dict{String, Any}) at main.jl:335
|                 run_cli_main(args::Vector{String}) at main.jl:442
|                 run_cli(args::Vector{String}) at common.jl:118
|                 run_cli() at common.jl:103
|                 top-level scope at baysor:3
|                 eval at boot.jl:373 [inlined]
|                 include_string(mapexpr::typeof(identity), mod::Module, code::String, filename::String) at loading.jl:1196
|                 _include(mapexpr::Function, mod::Module, _path::String) at loading.jl:1253
|                 include(mod::Module, _path::String) at Base.jl:420
|                 exec_options(opts::Base.JLOptions) at client.jl:292
|                 _start() at client.jl:495
└ Baysor /Users/xvk205/.julia/packages/Baysor/tda03/src/cli/common.jl:132

EDIT:

I've just tried running it without any prior segmentation (only providing the scale parameter) and I'm getting a similar error.

Command:
baysor run -x x_c -y y_c -z z_c --gene gene -o ./results_C1-2/ -i 500 --force-2d -p --n-clusters 7 -s 40 ../C1-2_results.txt

Output:

┌ Warning: No config file provided. Using default parameters.
└ @ Baysor ~/.julia/packages/Baysor/tda03/src/cli/main.jl:96
[14:55:31] Info: (2021-12-07) Run Baysor v0.5.1
[14:55:31] Info: Loading data...
[14:55:50] Info: Loaded 126550 transcripts
[14:55:51] Info: Estimating noise level
[14:55:57] Info: Done
[14:55:59] Info: Clustering molecules...
Progress: 100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| Time: 0:00:14
  Iteration:                  306
  Max. difference:            0.0010753232764250476
  Fraction of probs changed:  0.6586013433425524
[14:56:17] Info: Algorithm stopped after 306 iterations. Error: 0.00108. Converged: true.
[14:56:17] Info: Done
[14:56:19] Info: Initializing algorithm. Scale: 40.0, scale std: 10.0, initial #components: 84366, #molecules: 126550.
[14:56:21] Info: Using 2D coordinates
Progress: 100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| Time: 0:01:28
  Iteration:        500
  Noise level, %:   22.36
  Num. components:  1455
[14:57:55] Info: Processing complete.
[14:58:02] Info: Estimating local colors
[14:58:32] Info: Saving results to ./results_C1-2/segmentation.csv
[14:58:33] Info: Plot diagnostics
[14:59:38] Info: Estimating boundary polygons
[15:00:07] Error: MethodError(LightGraphs.kruskal_mst, ({144, 157} undirected simple Int64 graph with Float64 weights,), 0x0000000000007b62)
|
|                 border_mst(border_points::Vector{Vector{UInt32}}; min_edge_length::Float64, max_edge_length::Float64) at boundary_estimation.jl:65
|                 border_mst(border_points::Vector{Vector{UInt32}}) at boundary_estimation.jl:42
|                 _broadcast_getindex_evalf at broadcast.jl:670 [inlined]
|                 _broadcast_getindex at broadcast.jl:643 [inlined]
|                 getindex at broadcast.jl:597 [inlined]
|                 copy at broadcast.jl:943 [inlined]
|                 materialize at broadcast.jl:904 [inlined]
|                 extract_polygons_from_label_grid(grid_labels::Matrix{UInt16}; min_border_length::Int64, shape_method::Symbol, max_dev::Float64, exclude_labels::Vector{Int64}, offset::Vector{Float64}, grid_step::Float64) at boundary_estimation.jl:231
|                 (::Baysor.var"#extract_polygons_from_label_grid##kw")(::NamedTuple{(:min_border_length, :shape_method, :max_dev, :exclude_labels, :offset, :grid_step), Tuple{Int64, Symbol, Float64, Vector{Int64}, Vector{Float64}, Float64}}, ::typeof(Baysor.extract_polygons_from_label_grid), grid_labels::Matrix{UInt16}) at boundary_estimation.jl:223
|                 boundary_polygons(pos_data::Matrix{Float64}, cell_labels::Vector{Int64}; min_x::Nothing, max_x::Nothing, grid_step::Float64, min_border_length::Int64, shape_method::Symbol, max_dev::Float64, bandwidth::Float64, exclude_labels::Vector{Int64}, kwargs::Base.Pairs{Symbol, Union{}, Tuple{}, NamedTuple{(), Tuple{}}}) at boundary_estimation.jl:270
|                 boundary_polygons at boundary_estimation.jl:254 [inlined]
|                 #boundary_polygons#214 at boundary_estimation.jl:248 [inlined]
|                 (::Baysor.var"#boundary_polygons##kw")(::NamedTuple{(:grid_step, :bandwidth), Tuple{Float64, Float64}}, ::typeof(Baysor.boundary_polygons), spatial_df::DataFrames.DataFrame, args::Vector{Int64}) at boundary_estimation.jl:248
|                 plot_transcript_assignment_panel(df_res::DataFrames.DataFrame, assignment::Vector{Int64}, args::Dict{String, Any}; clusters::Vector{Int64}, prior_polygons::Vector{Matrix{Float64}}, gene_colors::Vector{ColorTypes.Lab{Float64}}) at main.jl:212
|                 (::Baysor.var"#plot_transcript_assignment_panel##kw")(::NamedTuple{(:clusters, :prior_polygons, :gene_colors), Tuple{Vector{Int64}, Vector{Matrix{Float64}}, Vector{ColorTypes.Lab{Float64}}}}, ::typeof(Baysor.plot_transcript_assignment_panel), df_res::DataFrames.DataFrame, assignment::Vector{Int64}, args::Dict{String, Any}) at main.jl:205
|                 save_segmentation_results(bm_data::Baysor.BmmData{2}, gene_names::Vector{String}, args::Dict{String, Any}; mol_clusts::NamedTuple{(:exprs, :assignment, :diffs, :assignment_probs, :change_fracs), Tuple{Matrix{Float64}, Vector{Int64}, Vector{Float64}, Matrix{Float64}, Vector{Float64}}}, comp_segs::Nothing, prior_polygons::Vector{Matrix{Float64}}) at main.jl:414
|                 (::Baysor.var"#save_segmentation_results##kw")(::NamedTuple{(:mol_clusts, :comp_segs, :prior_polygons), Tuple{NamedTuple{(:exprs, :assignment, :diffs, :assignment_probs, :change_fracs), Tuple{Matrix{Float64}, Vector{Int64}, Vector{Float64}, Matrix{Float64}, Vector{Float64}}}, Nothing, Vector{Matrix{Float64}}}}, ::typeof(Baysor.save_segmentation_results), bm_data::Baysor.BmmData{2}, gene_names::Vector{String}, args::Dict{String, Any}) at main.jl:394
|                 run_cli_main(args::Vector{String}) at main.jl:479
|                 run_cli(args::Vector{String}) at common.jl:118
|                 run_cli() at common.jl:103
|                 top-level scope at baysor:3
|                 eval at boot.jl:373 [inlined]
|                 include_string(mapexpr::typeof(identity), mod::Module, code::String, filename::String) at loading.jl:1196
|                 _include(mapexpr::Function, mod::Module, _path::String) at loading.jl:1253
|                 include(mod::Module, _path::String) at Base.jl:420
|                 exec_options(opts::Base.JLOptions) at client.jl:292
|                 _start() at client.jl:495
└ Baysor /Users/xvk205/.julia/packages/Baysor/tda03/src/cli/common.jl:132

[VPetukhov]

Hi @GMaciag ,
Probably it's not relevant anymore, but the new version of the package (v0.6) uses a different method for polygon estimation and shouldn't have this error.