Add files via upload

Feng-Research · Jun 26, 2023 · fa56fe3 · fa56fe3
1 parent 59e8d38
commit fa56fe3
Showing 2 changed files with 205 additions and 0 deletions.
diff --git a/SPADE_score/SPF.jl b/SPADE_score/SPF.jl
@@ -0,0 +1,133 @@
+include("decompositionW2.jl")
+include("Clique_sm.jl")
+include("mx_func.jl")
+include("StarW.jl")
+include("Filter_fast2.jl")
+include("h_scoreW.jl")
+include("INC3.jl")
+include("sparsification.jl")
+include("HyperNodes.jl")
+include("Mapping_fast.jl")
+include("Unmapping.jl")
+include("decompositionW2.jl")
+
+
+
+using SparseArrays
+using LinearAlgebra
+using Clustering
+using NearestNeighbors
+using Distances
+using Laplacians
+using Arpack
+using Statistics
+using DelimitedFiles
+using StatsBase
+using Laplacians#master
+using Random
+
+using PyCall
+
+function SPF(PyA, L, ICr)
+
+    m, n = PyA.shape
+    colPtr = Int[i+1 for i in PyArray(PyA."indptr")]
+    rowVal = Int[i+1 for i in PyArray(PyA."indices")]
+    nzVal = Vector{Float64}(PyArray(PyA."data"))
+    A = SparseMatrixCSC{Float64,Int64}(m, n, colPtr, rowVal, nzVal)
+
+    fdnz = findnz(tril(A, -1))
+
+    rr = fdnz[1]
+    cc = fdnz[2]
+    W = fdnz[3]
+
+    ar = Any[]
+
+    for ii = 1:length(rr)
+
+        push!(ar, sort([rr[ii], cc[ii]]))
+
+    end
+
+    ar_org = copy(ar)
+
+    RedR = 1
+
+    #@time ar, idx_mat, ar_mat, Emat = decompositionW2(ar, L, RedR, W)
+    @time ar, idx_mat, ar_mat, Emat = decompositionW2_fast(A, L)
+
+    avgN = mx_func(ar_org) / mx_func(ar)
+    println("-------------avgN = ", avgN)
+
+    Lmat = length(ar_mat)
+
+    NN = [1,Lmat]
+
+    arF, V = sparsification(NN, ar, idx_mat, ar_mat)
+
+    ## Adding inter-cluster edges
+    V = 1:mx_func(ar)
+
+    for ii = 1:length(idx_mat)
+
+    id1 = idx_mat[end-ii+1]
+
+        V = V[id1]
+
+    end
+
+    dict2 = Dict{Any, Any}()
+
+    for ii =1:length(ar_org)
+
+    nd1 = sort(ar_org[ii])
+
+    Vnd = sort(V[nd1])
+
+    if Vnd[1]!=Vnd[2]
+
+        vals = get!(Array{Int64,1},dict2, Vnd)
+
+        push!(vals, ii)
+
+    end
+
+    end # for ii
+
+    KS = collect(keys(dict2))
+
+    VL = collect(values(dict2))
+
+    EM1 = Emat[1]
+
+    for ii = 1:length(VL)
+
+        vl1 = VL[ii]
+
+        SP = sortperm(EM1[vl1])
+
+        TH = ceil(Int, length(SP) * ICr)
+
+        ES = vl1[SP[1:TH]]
+
+        append!(arF, ar_org[ES])
+
+    end
+
+    println("-------------Density = ", length(arF)/mx_func(arF))
+
+    AS = Clique_sm(arF)
+
+    # Import the required Python library
+    scipy_sparse = pyimport("scipy.sparse")
+    # Extract the components of the SparseMatrixCSC and adjust indices for 0-based Python
+    indptr = AS.colptr .- 1  # Subtract 1 for Python's 0-based indexing
+    indices = AS.rowval .- 1  # Subtract 1 for Python's 0-based indexing
+    data = AS.nzval
+    # Create a scipy.sparse.csr_matrix in Python directly using Julia arrays
+    PyAS = scipy_sparse.csr_matrix((data, indices, indptr), shape=(AS.m, AS.n))
+
+    return PyAS
+
+end
diff --git a/SPADE_score/eigen.jl b/SPADE_score/eigen.jl
@@ -0,0 +1,72 @@
+using LinearAlgebra
+using LinearMaps
+using MAT
+using SparseArrays
+using Arpack
+
+using PyCall, SparseArrays
+
+function scipyCSC_to_julia(A)
+    m, n = A.shape
+    colPtr = Int[i+1 for i in PyArray(A."indptr")]
+    rowVal = Int[i+1 for i in PyArray(A."indices")]
+    nzVal = Vector{Float64}(PyArray(A."data"))
+    B = SparseMatrixCSC{Float64,Int}(m, n, colPtr, rowVal, nzVal)
+    return PyCall.pyjlwrap_new(B)
+end
+
+
+function main(PyX, PyY, k::Int64)
+
+    rowval = Int[i+1 for i in PyArray(PyX."row")]
+    colval = Int[i+1 for i in PyArray(PyX."col")]
+    Val = Vector{Float64}(PyArray(PyX."data"))
+    X = sparse(rowval, colval, Val)
+
+    rowval = Int[i+1 for i in PyArray(PyY."row")]
+    colval = Int[i+1 for i in PyArray(PyY."col")]
+    Val = Vector{Float64}(PyArray(PyY."data"))
+    Y = sparse(rowval, colval, Val)
+
+
+
+    (Λ, V) = eigs(X, Y, nev=k, tol=1e-6, which=:LM)
+
+    return Λ, V
+end
+
+function not_main(PyX, k::Int64)
+
+    rowval = Int[i+1 for i in PyArray(PyX."row")]
+    colval = Int[i+1 for i in PyArray(PyX."col")]
+    Val = Vector{Float64}(PyArray(PyX."data"))
+    X = sparse(rowval, colval, Val)
+
+    (Λ, V) = eigs(X, nev=k, which=:SM,maxiter=500000)
+
+    idx = sortperm(abs.(Λ))
+    Λ_sorted = Λ[idx]
+    V_sorted = V[:, idx]
+
+    # Remove the first (smallest) eigenvalue and corresponding eigenvector
+    Λ_sorted = Λ_sorted[2:end]
+    V_sorted = V_sorted[:, 2:end]
+    return Λ_sorted, V_sorted
+end
+
+function plot_main(PyX, k::Int64)
+
+    rowval = Int[i+1 for i in PyArray(PyX."row")]
+    colval = Int[i+1 for i in PyArray(PyX."col")]
+    Val = Vector{Float64}(PyArray(PyX."data"))
+    X = sparse(rowval, colval, Val)
+
+    (Λ, V) = eigs(X, nev=k, which=:SM,maxiter=500000)
+
+    idx = sortperm(abs.(Λ))
+    Λ_sorted = Λ[idx]
+    V_sorted = V[:, idx]
+
+    return Λ_sorted, V_sorted
+end
+