add/test/doc counterpart of Python AnnData's obs_names_make_unique

docs/src/objects.md

@@ -68,6 +68,12 @@ Make duplicate `var_names` unique by appending a numbered suffix.
 var_names_make_unique!(ad)
 ```
 
+Similarly, we can make `obs_names` unique also.
+
+```@example 1
+obs_names_make_unique!(ad)
+```
+
 ## MuData
 
 The basic idea behind a multimodal object is _key_ ``\rightarrow`` _value_ relationship where _keys_ represent the unique names of individual modalities and _values_ are `AnnData` objects that contain the correposnding data. Similarly to `AnnData` objects, `MuData` objects can also contain rich multimodal annotations.

src/Muon.jl

@@ -15,7 +15,7 @@ using FileIO
 
 export readh5mu, readh5ad, writeh5mu, writeh5ad, isbacked, update_obs!, update_var!, update!
 export AnnData, MuData
-export var_names_make_unique!
+export var_names_make_unique!, obs_names_make_unique!
 
 import Pkg
 # this executes only during precompilation

src/util.jl

@@ -122,28 +122,42 @@ Make `A.var_names` unique by appending `join` and sequential numbers
 (1, 2, 3 etc) to duplicate elements, leaving the first unchanged.
 """
 function var_names_make_unique!(A::AnnData, join='-')
-    duplicates = duplicateindicies(A.var_names)
+    index_make_unique!(A.var_names, join)
+end
+
+"""
+    obs_names_make_unique!(A::AnnData, join = '-')
+
+Make `A.obs_names` unique by appending `join` and sequential numbers
+(1, 2, 3 etc) to duplicate elements, leaving the first unchanged.
+"""
+function obs_names_make_unique!(A::AnnData, join='-')
+    index_make_unique!(A.obs_names, join)
+end
+
+function index_make_unique!(index, join)
+    duplicates = duplicateindicies(index)
 
     if isempty(duplicates)
         @info "var names are already unique, doing nothing"
         return nothing
     end
 
     example_colliding_names = []
-    set = Set(A.var_names)
+    set = Set(index)
 
     for (name, positions) in duplicates
         i = 1
         for pos in Iterators.rest(positions, 2)
             while true
-                potential = string(A.var_names[pos], join, i)
+                potential = string(index[pos], join, i)
                 i += 1
                 if potential in set
                     if length(example_colliding_names) <= 5
                         push!(example_colliding_names, potential)
                     end
                 else
-                    A.var_names[pos] = potential
+                    index[pos] = potential
                     push!(set, potential)
                     break
                 end

test/anndata.jl

@@ -40,15 +40,24 @@ end
     test_ad_slicing(subad, 50, 5, x[i, 3:7])
 end
 
-@testset "anndata functions" begin
+@testset "unique names" begin
     @test_logs (:info,) var_names_make_unique!(ad)
+    @test_logs (:info,) obs_names_make_unique!(ad)
     ad2 = deepcopy(ad)
     ad2.var_names[3] == "10"
+    ad2.obs_names[90] == "obs_30"
     var_names_make_unique!(ad2)
+    obs_names_make_unique!(ad2)
     @test allunique(ad2.var_names)
+    @test allunique(ad2.obs_names)
     ad2.var_names[10] = "10-1"
     ad2.var_names[3] = "10"
     ad2.var_names[4] = "10"
+    ad2.obs_names[11] = "obs_10-1"
+    ad2.obs_names[10] = "obs_10"
+    ad2.obs_names[9] = "obs_10"
     @test_logs (:warn,) var_names_make_unique!(ad2)
+    @test_logs (:warn,) obs_names_make_unique!(ad2)
     @test allunique(ad2.var_names)
+    @test allunique(ad2.obs_names)
 end