Hi All,

I have a class called Mutant(), its constructor generates a random FASTA sequence.

A class method writeToPDB() uses modeller to push the FASTA into a template PDB. This is already working very well.

If I do 1000 mutants this takes an eternity (one after the other)... so I thought of crafting a ThreadPoolExecutor.

Modeller runs inside the class method. Concurrent calls should not share memory with other instances of Mutant(). 

mutants = [] #Holds an Array of mutant objects

executor = ThreadPoolExecutor(max_workers=20) #Thread executor

for i in range(1000):
    mutants.append(Mutant())
    executor.submit(mutants[i].writeToPDB, templatePDB)

This simply tells the executor to queue a new thread with the writeToPDB() method.

When I try to do this, it seems that modeller is being executed in the first call and sharing its state with all subsequent calls... so I get like 90 models (not 1000) written in whatever state modeller was when running the other instances.

How can I force a class method to use its own modeller instance?.

This is the method:

        env = environ()
        env.libs.topology.read(file='$(LIB)/top_heav.lib') 
        env.libs.parameters.read(file='$(LIB)/par.lib')

        aln = alignment(env)
        mdl = model(env, file=code)
        aln.append_model(mdl, atom_files=code, align_codes=code)
        residue = self.fasta[resid-Mutant.compensate]
        sel.mutate(residue_type=Mutant.res1to3[residue]) #mutate
      
        aln.append_model(mdl, align_codes='mut') 
        mdl.clear_topology()
        mdl.generate_topology(aln['mut'])
        mdl.transfer_xyz(aln) 
        mdl.build(initialize_xyz=False, build_method='INTERNAL_COORDINATES')

    
        name = str(self.pdbpath)+str(self.id)+".pdb"
        mdl.write(file=name)