search

thecailab / SPADE

GNU General Public License v3.0
0 stars 1 forks source link

Decrease computational cost #2

Open peicai opened 3 months ago

peicai commented 3 months ago

Hi, thank you very much for the package.

I tried to follow the provided scripts for identifying SV genes between groups. I am currently working on a dataset with approximately 4,000 spots and 13,743 genes in each group. However, I found that it takes almost 2 hours to fit just one gene.

Are there any parameters I could tune to run it faster, or do you have any recommendations for running the method more efficiently? Any suggestions would be greatly appreciated.

Thank you in advance. Best, Peiying

thecailab commented 3 months ago

Hi Peiying,

Thank you for being interested in our package. One of the solutions you can try is using binning strategy (combine a few cells into one) to decrease the number of samples in the image. You can find the function for binning online. If not, let us know. We can write a function and upload it to the package later.

Best, Fei

peicai commented 2 months ago

Hi Fei, thank you for your suggestion. I could use binning, but the dataset I am using is Visium data, where each spot already contains 1-10 cells. I believe handling 4000 spots should not be overly challenging. Is the high computational cost due to the large number of genes? Perhaps utilizing multiple cores to accelerate the function is an option?

best, Peiying