Merge pull request #3 from ChelseaTrotter/master

Create Binary for Genenetwork, working version for data input in Rqtl2 format
senresearch · Oct 28, 2020 · e9631ad · e9631ad
2 parents 499bb13 + 575bb93
commit e9631ad
Show file tree

Hide file tree

Showing 13 changed files with 24 additions and 285 deletions.
diff --git a/bin/MyApp/Project.toml b/bin/MyApp/Project.toml
diff --git a/bin/MyApp/precompile_app.jl b/bin/MyApp/precompile_app.jl
diff --git a/bin/MyApp/src/MyApp.jl b/bin/MyApp/src/MyApp.jl
diff --git a/bin/build-bin.jl b/bin/build-bin.jl
diff --git a/bin/install_packages.jl b/bin/install_packages.jl
diff --git a/r/Rqtl2scan.R b/r/Rqtl2scan.R
diff --git a/r/cleaning.R b/r/cleaning.R
diff --git a/sh/rqtl-julia.sh b/sh/rqtl-julia.sh
diff --git a/sh/run_lmgpu.sh b/sh/run_lmgpu.sh
diff --git a/src/LMGPU.jl b/src/LMGPU.jl
@@ -6,21 +6,19 @@ using LinearAlgebra
 using Base.Threads
 using ZipFile
 using CUDA
-# using CuArrays
-# using CUDAnative
-# using CUDAdrv
-# import CuArrays.CuArray
+
 
 #put all your source file here.
 include("data_io.jl")
+export get_geno_data, get_pheno_data, get_gmap_file
 include("util.jl")
 include("cpu.jl")
+export cpurun 
 include("gpu.jl")
 include("common_func.jl")
 include("match_gmap_info.jl")
+export  get_gmap_info, match_gmap
 # include("cli.jl")
 
-#put all your public functions (functions that you want user to use) here.
-export get_geno_data, get_pheno_data, cpurun, get_gmap_info, match_gmap #, gpurun
 
 end # module
diff --git a/src/cpu.jl b/src/cpu.jl
@@ -61,7 +61,7 @@ end
 
 
 ##################### Running CPU Function ###################
-function cpurun(a::AbstractArray{<:Real, 2}, b::AbstractArray{<:Real, 2}, n::Int, maxlod::Bool)
+function cpurun(a::AbstractArray{<:Real, 2}, b::AbstractArray{<:Real, 2}, n::Int, export_matrix::Bool)
     a_std = get_standardized_matrix(a);
     b_std = get_standardized_matrix(b);
     #step 2: calculate R, matrix of corelation coefficients
@@ -70,8 +70,9 @@ function cpurun(a::AbstractArray{<:Real, 2}, b::AbstractArray{<:Real, 2}, n::Int
     # lod = lod_score(n, r);
     lod = lod_score_multithread(n,r)
 
-    if maxlod 
-        println("exporting max lod")
+
+    if !export_matrix 
+        println("Calculating max lod")
         return find_max_idx_value(lod)
     else 
         println("exporting matrix.")

diff --git a/src/data_io.jl b/src/data_io.jl
@@ -12,13 +12,24 @@ end
 #     return convert(Array{datatype,2}, pheno)
 # end
 
+function try_string2num(num)
+    return tryparse(Float64,num) != nothing
+end
+
 
 function get_pheno_data(file, datatype; transposed=true)
 
-    #first column is individual ID such as : BXD1
-    pheno = readdlm(file, ','; skipstart=1)
+    #first column is individual ID such as : BXD1 , need to be removed. 
+    pheno = readdlm(file, ','; skipstart=1)[:, 2:end]
+
+    if pheno[1,end] == "f" || pheno[1,end] == "m"
+        @info "Removing sex column of phenotype. "
+        pheno = pheno[:, 1:end-1]
+
+    end
+
     pheno = convert(Array{datatype,2}, pheno)
-    # pheno = convert2float.(pheno, datatype)
+
     if transposed 
         return transpose(pheno) |> collect
     else 

diff --git a/src/match_gmap_info.jl b/src/match_gmap_info.jl
@@ -16,24 +16,8 @@ function extension(url::String)
     end
 end
 
-function get_gmap_info(rqtl_file)
-
-    # if passing in rqtl_file as a zip, extract gmap file.
-    if extension(rqtl_file) == ".zip"
-        dir = ZipFile.Reader(rqtl_file)
-        f = findfile(dir, "gmap.csv")
-        gmap = readdlm(f, ',')
-        close(dir)
-    # if passing in just gmap file.
-    elseif extension(rqtl_file) == ".csv"
-        if occursin("gmap.csv", rqtl_file)
-            gmap = readdlm(rqtl_file, ',')
-        else
-            error("no gmap file found.")
-        end
-    else
-        error("Rqtl file is not passed in as a .zip, need to handle this.")
-    end
+function get_gmap_info(gmap_file)
+    gmap = readdlm(gmap_file, ',', header=true)
     return gmap
 
 end