Get stuck on 'Estimating local colors', memory use increases and freeze

[HaojiaWu]

Hi @VPetukhov I updated my baysor version to 0.6.1 and ran it on the example data from startmap. I got stuck on the 'Estimating local colors' step. The memory starts to increases indefinitely and the process freezes. Can you take a look what might cause this?

[07:25:37] Info: Run Rd935f0af5
[07:25:37] Info: (2023-07-18) Run Baysor v0.6.1
[07:25:37] Info: Loading data...
[07:25:41] Info: Loaded 949505 transcripts
[07:25:44] Info: Estimating noise level
[07:26:00] Info: Done
[07:26:22] Info: Clustering molecules...
[07:26:27] Warning: ICA did not converge, fall back to random initialization
└ Baysor.Processing /home/haojiawu/.julia/packages/Baysor/L94P7/src/processing/bmm_algorithm/molecule_clustering.jl:147
[07:28:28] Info: Algorithm stopped after 509 iterations. Error: 0.00232. Converged: true.
[07:28:28] Info: Done
[07:28:28] Info: Initializing algorithm. Scale: 100.0, scale std: 25.0, initial #components: 27128, #molecules: 949505.
[07:28:33] Info: Using the following additional information about molecules: [:confidence, :cluster]
[07:28:33] Info: Using 3D coordinates
[07:52:00] Info: Processing complete.
[07:52:18] Info: Estimating local colors

Here is the script I used.

baysor  run -x x -y y -z z -g gene -s 100 -m 70 -o ./starmap -p --save-polygons=geojson ./data/molecules.csv

Thanks!

[VPetukhov]

That looks like the same problem as https://github.com/kharchenkolab/Baysor/issues/78 . Will be fixed in the next release.

[VPetukhov]

Done.