xzz-1995
commented
3 months ago Hi Che, For the choice of clustering method, we use random tree to ensure that every cell partitioned has sufficient evidence of tumor heterogeneity. while leiden is widely used in the network analysis of large-scale datasets. For spatial data, usually there are about 3000–4000 spots in each sample, so with more cores used in infercnv, the time would not be too long. Regarding the code problem, it was actually a bug, which I have corrected in the script. Thank you for reminding me.
Dear Developers,
Firstly, I would like to extend my sincere gratitude for your efforts in developing the Cottrazm package. It has been instrumental in my research, and I appreciate the work that has gone into it.
I am reaching out to seek your advice regarding an issue I encountered while using the random_trees method within the package. I've noticed that this method significantly increases the computation time. I am curious if the use of random_trees is an absolute necessity? Have you, by any chance, experimented with the default leiden method from infercnv for the identification of core malignant spots? If so, I would be grateful to learn about the outcomes of such an approach.
Additionally, while reviewing the STCNVScore.R script, I stumbled upon a discrepancy in the naming convention and strategy used in the comments as opposed to the actual code implementation. Could you kindly clarify how this section of the code is intended to be utilized in practice?
Your insights and clarifications on these matters would be immensely helpful for me and perhaps others in the community facing similar challenges.
Thank you once again for your valuable contribution to the field.
Best regards, Che