search

kharchenkolab / Baysor

Bayesian Segmentation of Spatial Transcriptomics Data
MIT License
142 stars 29 forks source link

Baysor run stuck after 2D coordinates #84

Closed markutz556 closed 11 months ago

markutz556 commented 1 year ago

Hi,

I tried to use Baysor on my data and I think it will be very helpful with the analysis.

I'm running it with my mac (16GB memory with an Intel chip) and I'm stuck with the following step. I've been waiting for over 4 hours and I'm not sure if it's still running or there is any issues.

Image 2023-7-17 at 17 22

I'm using Baysor version 0.6.1 (install as a Julia package). I wonder if my mac is capable of running Baysor with 1 million transcripts (or if it's possible to run with some modifications).

Any ideas would be appreciated.

Thanks, Mark

VPetukhov commented 12 months ago

Hi @markutz556 , That seems to be the same problem of incompatibility with julia 1.9, as https://github.com/kharchenkolab/Baysor/issues/78 . Fixed it in the last develop commit. Hopefully, will release it within a week.

markutz556 commented 12 months ago

Hi @VPetukhov,

Thanks for your response. I found that this can be solved if I used the --no-ncv-estimation flag. I wonder if the result would be very different with --no-ncv-estimation flag compared with the run without this flag.

And maybe a naive question. What are the major differences between with ncv estimation and without? Thanks.

Best, Mark

VPetukhov commented 11 months ago

@markutz556 , it's all about plotting. Without NCVs you won't get a segmentation plot, and your output won't have ncv_color column that could be used for molecule data visualization.

VPetukhov commented 11 months ago

The release is ready now, so it should be fixed.