Seyoon Ko
Seyoon Ko
@BrendonChau While I haven't had a chance to look into it yet, that certainly makes sense to me. The previous example wasn't just "tall" enough--perhaps we can change behavior in...
Which one do you think more affects the performance, parallelizing over the tiles in a column or loop unrolling via `Base.Cartesian.@nexprs`?
I need some clarification, as there already is parallelization happening with the current package. So do you mean limiting the parallelization to a single column has more impact than 20~30%...
I don't think it's anybody's fault, Ben. It's just that we have not designed a kernel for this specific operation. `LoopVectorization` is not really a good solution for SparseMatrixCSC. For...
The function `Base.convert(::Type{T}, x::UInt8, ::Union{Val{1}, Val{2}, Val{3}})` and `Base.copyto!( v::AbstractVecOrMat{T}, s::AbstractSnpArray; model::Union{Val{1}, Val{2}, Val{3}} = ADDITIVE_MODEL, center::Bool = false, scale::Bool = false, impute::Bool = false ) ` are only defined...
Changing ```julia Array{T,N}(s::AbstractSnpArray; kwargs...) where {T,N} = copyto!(Array{T,N}(undef, size(s)), s; kwargs...) ``` on https://github.com/OpenMendel/SnpArrays.jl/blob/master/src/snparray.jl#L370 to something like ```julia Array{T,N}(s::AbstractSnpArray; kwargs...) where {T,N} = throw(ArgumentError("Cannot convert $(typeof(s)) to Array{$T,$N}")) Array{T,N}(s::AbstractSnpArray; kwargs...)...
In Plink 2's fixed-width format, 0x00 encodes 0, 0x01 encodes 1, 0x02 encodes 2, and 0x03 encodes missing. This might be a good Int-representation for a SnpArray.
These are the first steps to set up gitlab-ci for GPU code with Julia: https://github.com/JuliaGPU/gitlab-ci > First of all, you need to be the owner of your repository or Gitlab...
The source code is here. https://github.com/chrchang/plink-ng/blob/master/2.0/plink2_ld.cc ... but can we find a text description?
Looks relevant: https://gsejournal.biomedcentral.com/articles/10.1186/s12711-018-0404-z