ValueError

[CaiYitao]

Got the following error, what could be the problem?
the functions are from rdkit

@py begin
	import rdkit
	import rdkit.Chem.AllChem as AC
	import rdkit.Chem as Chem
	import numpy as np
	end

or use julia array:

Thank you!

[icweaver]

I think this usually happens when trying to do a py->jl conversion that is not supported. For example:

julia> using PythonCall

julia> Float64[pyfloat(4)]
ERROR: MethodError: Cannot `convert` an object of type Py to an object of type Float64
Closest candidates are:
  convert(::Type{T}, ::T) where T<:Number at ~/julia-1.7.2/share/julia/base/number.jl:6
  convert(::Type{T}, ::Number) where T<:Number at ~/julia-1.7.2/share/julia/base/number.jl:7
  convert(::Type{T}, ::Base.TwicePrecision) where T<:Number at ~/julia-1.7.2/share/julia/base/twiceprecision.jl:262
  ...
Stacktrace:
 [1] setindex!(A::Vector{Float64}, x::Py, i1::Int64)
   @ Base ./array.jl:903
 [2] getindex(#unused#::Type{Float64}, x::Py)
   @ Base ./array.jl:412
 [3] top-level scope
   @ REPL[2]:1

So I think a Julia object might be sneaking into your setup somewhere before it is passed to a python function that does not know how to handle it.

or use julia array

Same thing here. rdkit seems to be expecting a "Numeric array" (is this their lingo for a numpy array?), so passing the Julia array arr = zeros(1) instead of the numpy array arr = np.zeros(1) will cause ConvertToNumpyArray to rightly complain

[icweaver]

fwiw, trying a version of the example you shared seems to work alright for me at least:

Notebook file

### A Pluto.jl notebook ###
# v0.19.0

using Markdown
using InteractiveUtils

# ╔═╡ eb585990-b821-11ec-26ae-e928d6ac0b1d
begin
   using Pkg
   Pkg.activate(Base.current_project())
   
   using PythonCall
end

# ╔═╡ 0e7d0578-1e8f-4864-8047-50f3e8fb6952
@py begin
   import rdkit
   import rdkit.Chem.AllChem as AC
   import rdkit.Chem as Chem
   import numpy as np
end

# ╔═╡ bf1e5c85-3648-456e-8327-343a7c9df595
mol = Chem.MolFromSmiles("C1CCCOC1")

# ╔═╡ 51d28116-5823-4dad-a26c-8b727dba12f1
fp_vec = AC.GetMorganFingerprintAsBitVect(mol, radius=3, nBits=8)

# ╔═╡ 66288740-497b-454c-a6d9-b8c5e92141a8
begin
   arr = np.zeros(1)
   # I guess this mutates
   Chem.DataStructs.ConvertToNumpyArray(fp_vec, arr)
   arr
end

# ╔═╡ 51991470-a8fd-4e10-ae5a-e7c0a0b06ccf
print(fp_vec)

# ╔═╡ Cell order:
# ╠═eb585990-b821-11ec-26ae-e928d6ac0b1d
# ╠═0e7d0578-1e8f-4864-8047-50f3e8fb6952
# ╠═bf1e5c85-3648-456e-8327-343a7c9df595
# ╠═51d28116-5823-4dad-a26c-8b727dba12f1
# ╠═66288740-497b-454c-a6d9-b8c5e92141a8
# ╠═51991470-a8fd-4e10-ae5a-e7c0a0b06ccf

I don't know anything about chem, or what global variables are being passed in your screenshot, so I could be misunderstanding what is causing your issue here

[CaiYitao]

This is amazing! Thank you very much for your help!
Now I found that when we use for loop it spits that error, if we do not use for loop, it just works as you showed.

[cjdoris]

Your first error was on line 8, the one below the one highlighted, namely fps[i] = arr, which is complaining that you're trying to set an element of an array of floats to a Python object.

[CaiYitao]

Yeah, Thank you for pointing out, yeah, the a was mutated to an array of shapes (1, L), now this is interesting Python can do this

After I convert the python array into julia type Array, it gave this error

but after I change the

	Lₚ = 1024
	fps = np.zeros((length(y), Lₚ))
end

it works fine

Really appreciate your help!!!

[icweaver]

Your first error was on line 8, the one below the one highlighted, namely fps[i] = arr

oop, missed that line when transcribing the screenshot. Thanks @cjdoris!