Hallucination based protein design method with Alphafold2 as an oracle and SOLeNNoID discriminator network to produce solenoid proteins.
🚧 Work in progress.. 🚧
git clone https://github.com/yourusername/InSilicoEvolution.git cd InSilicoEvolutionpython3 src/main.py \ --parent_dir /path/to/output \ --population_size 20 \ --rounds 50 \ --solenoid_type alphabeta| Argument | Description | Default |
|---|---|---|
--parent_dir | Root directory for input/output | . |
--input_dir | Input FASTA folder name | in_silico_evolution_input |
--output_dir | ColabFold output folder name | in_silico_evolution_output |
--final_output_dir | Where final results are stored | output_statistics |
--num_repeats | Repeats of the sequence in FASTA | 6 |
--population_size | Genetic algorithm population size | 10 |
--parent_strategy | Parent selection strategy | wright-fisher |
--beta | Mutation strength parameter | 0.1 |
--children_proportion | Proportion of children per generation | 0.8 |
--rounds | Number of design rounds | 30 |
--sequences_batch_size | Number of sequences processed per batch | 1 |
--model_queries_per_batch | Number of queries per generation | 30 |
--starting_sequence | Provide a starting sequence | "" (auto-generated) |
--sequence_length | Length of generated starting sequence | 30 |
--min_solenoid | Threshold for solenoid confidence | 0.6 |
--min_plddt | Threshold for pLDDT confidence | 0.7 |
--solenoid_type | Solenoid class to target (beta, alphabeta, alpha) | beta |