Getting Cluster labels

[germanb27]

Description of feature

I was trying to get the cluster assigned to each cell but I think this feature is either not available or not easily accesible (I have tried with BaseClusterEstimator but can't do it). I wonder whether it could be implemented in order to add a column to my dataframe with the cluster.

[berombau]

Hi,

A complete example with the FlowSOM object would be something like this. The labels are in the column metaclustering:

In [1]: # Import the FlowSOM package
   ...: import flowsom as fs
   ...: 
   ...: # Load the FCS file
   ...: ff = fs.io.read_FCS("./tests/data/ff.fcs")
   ...: 
   ...: # Run the FlowSOM algorithm
   ...: fsom = fs.FlowSOM(
   ...:     ff, cols_to_use=[8, 11, 13, 14, 15, 16, 17], xdim=10, ydim=10, n_clusters=10, seed=42
   ...: )

In [8]: fsom.get_cell_data().obs
Out[8]: 
       clustering  distance_to_bmu  metaclustering
0              91         0.511422               0
1              95         1.114001               2
2              16         1.093650               1
3              55         0.953404               7
4              49         0.688398               3
...           ...              ...             ...
19220          46         0.675527               3
19221          67         0.882528               3
19222          11         0.476845               0
19223           0         0.573982               0
19224          61         0.593200               0

[19225 rows x 3 columns]

For the functional AnnData workflow with flowsom_clustering, it looks something like this. The labels are in the column FlowSOM_metaclusters.

In [10]: ff_clustered = fs.flowsom_clustering(ff, cols_to_use=[8, 11, 13, 14, 15, 16, 17], xdim=10, ydim=10, n_clusters=10, seed=42)
    ...: ff_clustered
2024-11-25 11:30:44.034 | DEBUG    | flowsom.main:__init__:84 - Reading input.
2024-11-25 11:30:44.035 | DEBUG    | flowsom.main:__init__:86 - Fitting model: clustering and metaclustering.
2024-11-25 11:30:44.194 | DEBUG    | flowsom.main:__init__:88 - Updating derived values.
Out[10]: 
AnnData object with n_obs × n_vars = 19225 × 18
    obs: 'clustering', 'distance_to_bmu', 'metaclustering', 'FlowSOM_clusters', 'FlowSOM_metaclusters'
    var: 'n', 'channel', 'marker', '$PnB', '$PnE', '$PnG', '$PnR', '$PnV', 'pretty_colnames', 'markers', 'channels', 'cols_used'
    uns: 'meta', 'n_nodes', 'n_metaclusters', 'FlowSOM'

In [11]: ff_clustered.obs
Out[11]: 
       clustering  distance_to_bmu  metaclustering  FlowSOM_clusters  FlowSOM_metaclusters
0              91         0.511422               0                91                     0
1              95         1.114001               2                95                     2
2              16         1.093650               1                16                     1
3              55         0.953404               7                55                     7
4              49         0.688398               3                49                     3
...           ...              ...             ...               ...                   ...
19220          46         0.675527               3                46                     3
19221          67         0.882528               3                67                     3
19222          11         0.476845               0                11                     0
19223           0         0.573982               0                 0                     0
19224          61         0.593200               0                61                     0

[19225 rows x 5 columns]

Should the .obs functionality be unclear, see the AnnData documentation.