removing labels from features

Author: racinmat

src/Mutagenesis/Mutagenesis.jl

@@ -85,12 +85,13 @@ function load_data(dir)
     metadata = read_data(metadata_path)
     labelkey = metadata["label"]
     targets = map(i -> i[labelkey], samples)
+    samples_without_label = map(x->delete!(copy(x), Symbol(labelkey)), samples)
     val_num = metadata["val_samples"]
     test_num = metadata["test_samples"]
     train_idxs = 1:length(samples)-val_num-test_num
     val_idxs = length(samples)-val_num-test_num+1:length(samples)-test_num
     test_idxs = length(samples)-test_num+1:length(samples)
-    samples, targets, train_idxs, val_idxs, test_idxs
+    samples_without_label, targets, train_idxs, val_idxs, test_idxs
 end
 
 read_data(path) = open(JSON3.read, path)

test/tst_mutagenesis.jl

@@ -10,4 +10,8 @@ end
     @test length(train_x) == length(train_y) == 100
     @test length(test_x) == length(test_y) == 44
     @test length(val_x) == length(val_y) == 44
+    # test that label is not contained in features
+    @test !any(haskey.(train_x, :mutagenic))
+    @test !any(haskey.(test_x, :mutagenic))
+    @test !any(haskey.(val_x, :mutagenic))
 end