Conformer generation isn't deterministic

[sevyharris]

When you create a new AutoTST species, it gets the geometry from RDKit's Embed functionality without using a random seed, so you get a different result each time.

Here are 3 different versions of butane AutoTST gives you if create the species 3 times.

import ase
import ase.visualize
import autotst.species

my_ase_atoms = []
for i in range(3):
    my_species = autotst.species.Species(smiles=['CCCC'])
    my_ase_atoms.append(my_species.conformers['CCCC'][0].get_ase_mol())
    
ase.visualize.view(my_ase_atoms[0] + my_ase_atoms[1] + my_ase_atoms[2], viewer='x3d')

This is a problem if you have a workflow that has to rebuild AutoTST objects every time it restarts. We should set the random seed.

[sevyharris]

Here's the code and result for the Chiral example (in the set_chirality() function):

# need chiral center example
all_atoms = ase.Atoms()
for i in range(10):

    my_chiral_smiles = 'ClC=C(C(Cl)N)O'
    my_chiral_species = autotst.species.Species(smiles=[my_chiral_smiles])
    my_chiral_ase_atoms.append(my_chiral_species.conformers[my_chiral_smiles][0].get_ase_mol())
    
    chiral_centers = my_chiral_species.conformers[my_chiral_smiles][0].get_chiral_centers()
    my_chiral_species.conformers[my_chiral_smiles][0].set_chirality(chiral_centers[0].index, 'R')
    
    all_atoms += my_chiral_species.conformers[my_chiral_smiles][0].get_ase_mol()
  

ase.visualize.view(all_atoms, viewer='x3d')

[sevyharris]

Example with TS class (this is after I've added the random seed for species conformers and chirality):

all_atoms = ase.Atoms()
my_reaction_smiles = 'CCCC+[OH]_O+[CH2]CCC'
for i in range(5):
    my_reaction = autotst.reaction.Reaction(label=my_reaction_smiles)
    my_reaction.ts

    all_atoms += my_reaction.ts['forward'][0].ase_molecule