R meet fatal error when use SILMLR package

[wwn7268]

Hi SIMLR team,

I am a student from China. I am using your R packages "SIMLR" to analysis single-cell RNA-seq data. I need to use the function "SIMLR_Large_Scale" to cluster big dataset. For some datasets, it runs normally. However, for others it may meet fetal error when running t-sne and terminate R. I want to know what happened and how to solve it. Thank you very much!

Best regard,
Tommy

[danro9685]

Dear @wwn7268,

I believe this may be a memory issue due to very big datasets or reduced memory where you are running SIMLR.

But to give you more feedbacks, I would need to know something more about when it gives you fatal error.

Best,
Daniele

[wwn7268]

For example,when I running a dataset with 23000 genes and 1000 cells using "SIMLR_Large_Scale". It can run it correctly before t-SNE, but terminate after it star the step "performing t-SNE".

[danro9685]

Dear @wwn7268,

I will need to be able to reproduce this in order to help you. Are you using the R package? One think you may try is to use MATLAB implementation.

Best,
Daniele

[yulijia]

I meet the same problem even using the latest package 1.18.0
@danro9685 would you mind providing a parameter officially in the R package to get rid of the TSNE step?

if (calculate_TSNE==FALSE){
do not calculate TSNE, don't return ydata
}else{
calculate TSNE, and return ydata
}

After having a look at SIMLR_Large_Scale and SIMLR function, I find you use different TSNE function (Rtsne for SIMLR_Large_Scale, tsne for the SIMLR), is this because you want to make sure it doesn't take a lot of time to calculate TSNE value for large size of data?