Properly display missing values for SnpLinAlg when impute=true (#88)

* Drop dependency on GeneticVariation.jl in favor of VariantCallFormat * proper display of getindex for missing data in SnpLinAlg * added some tests for missing data in snplinalg
OpenMendel · Jun 1, 2021 · 68a99b8 · 68a99b8
1 parent 2a7976e
commit 68a99b8
Show file tree

Hide file tree

Showing 2 changed files with 9 additions and 2 deletions.
diff --git a/src/linalg_direct.jl b/src/linalg_direct.jl
@@ -94,6 +94,7 @@ eltype(bm::SnpLinAlg) = eltype(bm.μ)
 
 function Base.getindex(s::SnpLinAlg{T}, i::Int, j::Int) where T
     x = SnpArrays.convert(T, getindex(s.s, i, j), s.model)
+    s.impute && isnan(x) && return s.μ[j]
     s.center && (x -= s.μ[j])
     s.scale && (x *= s.σinv[j])
     return x

diff --git a/test/runtests.jl b/test/runtests.jl
@@ -294,9 +294,15 @@ end
 
 @testset "copyto SnpLinAlg" begin
 for model in [ADDITIVE_MODEL, DOMINANT_MODEL, RECESSIVE_MODEL], t in [Float32, Float64]
-    mousela = SnpLinAlg{t}(mouse, model=ADDITIVE_MODEL, center=true, scale=true)
-    v = copyto!(zeros(1), @view(mousela[702]))
+    # imputing missing data
+    mousela = SnpLinAlg{t}(mouse, model=ADDITIVE_MODEL, center=true, scale=true, impute=true)
+    v = copyto!(zeros(1), @view(mousela[702])) # missing entry
+    @test isapprox(v[1], 1.113003134727478, atol=1e-6)
+    # not imputing missing data
+    mousela = SnpLinAlg{t}(mouse, model=ADDITIVE_MODEL, center=true, scale=true, impute=false)
+    v = copyto!(zeros(1), @view(mousela[702])) # missing entry
     @test isnan(v[1])
+    # no missing data
     EURla = SnpLinAlg{t}(EUR, model=model, center=true, scale=true)
     EURtrue = convert(Matrix{t}, EUR, model=model, center=true, scale=true)
     EURtest = copyto!(zeros(t, size(EUR)), EURla)