Dear Granja
It is quite a smart idea to apply the 'uwot_transform' to in situ projection of new data (disease sample of both single cell or bulk datasets) to a already calculated umap graph. This method, which is named 'LSI_projection' is exemplified in your script scRNA_02_Cluster_Disease_w_Reference_v1.R. However, when loading the saved umap model file from a tar file in 'scRNA-Projection-UMAP.zip', it complained for the unloaded nn_index. I looked into this issue further by run the load_uwot() code step by step and found a error message "Index size is not a multiple of vector size: Success (0)". and if applied model$nn_index$getNTrees() , I got zero. But , the nn1 file is not empty. As I described above, it seems that either the nn_index is empty or is not compatible with my R enviroment.
Help!
Meijiao
Dear Granja
It is quite a smart idea to apply the 'uwot_transform' to in situ projection of new data (disease sample of both single cell or bulk datasets) to a already calculated umap graph. This method, which is named 'LSI_projection' is exemplified in your script scRNA_02_Cluster_Disease_w_Reference_v1.R. However, when loading the saved umap model file from a tar file in 'scRNA-Projection-UMAP.zip', it complained for the unloaded nn_index. I looked into this issue further by run the load_uwot() code step by step and found a error message "Index size is not a multiple of vector size: Success (0)". and if applied model$nn_index$getNTrees() , I got zero. But , the nn1 file is not empty. As I described above, it seems that either the nn_index is empty or is not compatible with my R enviroment.
Help!
Meijiao