FindClusters is taking too long for ~700K cells

[buutrg]

Hi Seurat developers, I am using add Cluster on a 700K cells dataset, and it froze after Running Louvain algorithm for 12 hours. I have tried using default parameters with resolution = 1.5 and 1.8

Computing nearest neighbor graph
Computing SNN
Modularity Optimizer version 1.3.0 by Ludo Waltman and Nees Jan van Eck

Number of nodes: 694117
Number of edges: 8990440

Running Louvain algorithm...
0%   10   20   30   40   50   60   70   80   90   100%
[----|----|----|----|----|----|----|----|----|----|
**************************************************|
Maximum modularity in 10 random starts: 0.9372
Number of communities: 11279
Elapsed time: 249 seconds

Is this a normal behavior? Do you know how to solve this?

Thank you!

[sunhang526]

I have the similar problem now, have you solved it? please

[lizchcase]

Not sure if you already found your way through this but 700k is a ton of cells! This is great experimentally but will require a lot of computational resources. The amount of time it would take depends a lot on the specs of the computer/cluster you're running on. Generally, running with more nodes will make this go faster, and you will likely require quite a bit of memory/RAM

[xinglei01]

"Hello developers, I have a question. When the number of cells reaches over a million, the running speed of FindClusters with default parameters is too slow. Can I set singleton=F? What is the essential purpose of this parameter?"

[baike687]

I also have the same problem for about 500 000 cells, and It stays there for so long. Is it normal?

Modularity Optimizer version 1.3.0 by Ludo Waltman and Nees Jan van Eck

Number of nodes: 539830
Number of edges: 9682376

Running Louvain algorithm...
0%   10   20   30   40   50   60   70   80   90   100%
[----|----|----|----|----|----|----|----|----|----|
**************************************************|
Maximum modularity in 10 random starts: 0.9351
Number of communities: 2449
Elapsed time: 761 seconds