rndsmi.py

import selfies, random, sys

N = 1000
M = 500
selfies.set_semantic_constraints("octet_rule")
alphabet= list(selfies.get_semantic_robust_alphabet())
alphabet = list(filter(lambda x: '+' not in x and '-' not in x, alphabet))

rnd_smi = set()
while len(rnd_smi) < N:
    slf = ''.join(random.choices(alphabet, k=M))
    smi = selfies.decoder(slf)
    if len(smi) > 40:
        rnd_smi.add(smi)
    sys.stdout.write(f"\rmade {len(rnd_smi)} SMILES")
    sys.stdout.flush()
    
o  =open('rnd.smi', 'w')
for i in rnd_smi:
    o.write(f"{i}\n")