Mobius

A python package for optimizing peptide sequences using Bayesian optimization (BO).

Features

• Single or multi-objectives optimization, with constraints
• Supports linear and non-linear peptide sequences (macrocyclic, lasso, branched, etc, ...)
• Supports any non-standard amino-acid residues and modifications
• Use design protocols to customize the sequence optimization
• Integrate protein or chemical Language Models, and fine-tune them on existing data (classical or LoRA).
• Easily extensible to add your own molecular representations (fingerprints, graph kernels, GNN, ..)
• Can be seamlessly integrated with other tools (docking, pyRosetta, Damietta, AlphaFold, ..)

Examples

The quickiest way to get started with mobius is to look through the different examples. This notebooks will introduce you to the different features of mobius and give you some ideas how to integrate it in your own projects.

• Would BO have helped me?: Colab notebook
• Optimization of linear peptides towards MHC class I: Jupyter notebook
• Example of multi-objectives optimization: Jupyter notebook
• Integration of non-standard amino acids and macrocyclic peptides: Jupyter notebook (tutorial) and Jupyter notebook (thrombin)
• Protein language models for peptide optimization: Jupyter notebook
• Example of peptide permeability optimization: Jupyter notebook
• Think of proteins as... kind of a big peptide: Jupyter notebook GFP and Jupyter notebook GB1
• Graph Neural Networks and Graph Kernel for peptide optimization: Jupyter notebook (GNN) and Jupyter notebook (GK)
• Use my own molecular representation: Jupyter notebook

Documentation

The installation instructions, documentation and tutorials can be found at mobius.readthedocs.io.

Installation

First, you need to download the python package:

git clone https://git.scicore.unibas.ch/schwede/mobius.git

Now you can set up the environment using mamba. For more instructions on how to install mamba, see: https://github.com/conda-forge/miniforge#miniforge3

cd mobius
mamba env create -f environment.yaml -n mobius
mamba activate mobius

Once you created the environment, you can install the package using the following command:

pip install -e .

Quick tutorial

import numpy as np
from mobius import Map4Fingerprint
from mobius import GPModel, ExpectedImprovement, TanimotoSimilarityKernel
from mobius import LinearPeptideEmulator
from mobius import homolog_scanning
from mobius import convert_FASTA_to_HELM
from mobius import Planner, SequenceGA

Simple linear peptide emulator/oracle for MHC class I A*0201. The Position Specific Scoring Matrices (PSSM) can be downloaded from the IEDB database (see Scoring matrices of SMM and SMMPMBEC section). WARNING: This is for benchmarking purpose only. This step should be an actual lab experiment.

lpe = LinearPeptideEmulator('IEDB_MHC/smmpmbec_matrix/HLA-A-02:01-9.txt',)

Now we define a peptide sequence we want to optimize:

lead_peptide = convert_FASTA_to_HELM('HMTEVVRRC')[0]

Then we generate the first seed library of 96 peptides using a homolog scanning sequence-based strategy.

seed_library = [lead_peptide]

for seq in homolog_scanning(lead_peptide):
    seed_library.append(seq)

    if len(seed_library) >= 96:
        print('Reach max. number of peptides allowed.')
        break

The seed library is then virtually tested (Make/Test) using the linear peptide emulator we defined earlier. WARNING: This is for benchmarking purpose only. This step is supposed to be an actual lab experiment.

pic50_seed_library = lpe.score(seed_library)

Once we have the results from our first lab experiment we can now start the Bayesian Optimization (BO). First, we define the molecular fingerprint we want to use as well as the surrogate model (Gaussian Process) and the acquisition function (Expected Improvement).

map4 = Map4Fingerprint(input_type='helm', dimensions=4096, radius=1)
gpmodel = GPModel(kernel=TanimotoSimilarityKernel(), transform=map4)
acq_fun = ExpectedImprovement(gpmodel, maximize=False)

... and now let's define the search protocol in a YAML configuration file (design_protocol.yaml) that will be used to optimize the peptide sequence. This YAML configuration file defines the design protocol, in which you need to define the peptide scaffold, linear here. Additionnaly, you can specify the sets of monomers to be used at specific positions during the optimization. You can also define some filtering criteria to remove peptide sequences that might exhibit some problematic properties during synthesis, such as self-aggregation or solubility.

design:
  monomers: 
    default: [A, C, D, E, F, G, H, I, K, L, M, N, P, Q, R, S, T, V, W, Y]
    APOLAR: [A, F, G, I, L, P, V, W]
    POLAR: [C, D, E, H, K, N, Q, R, K, S, T, M]
    AROMATIC: [F, H, W, Y]
    POS_CHARGED: [K, R]
    NEG_CHARGED: [D, E]
  polymers:
    - PEPTIDE1{X.M.X.X.X.X.X.X.X}$$$$V2.0:
        PEPTIDE1:
          1: [AROMATIC, NEG_CHARGED]
          4: POLAR
          9: [A, V, I, L, M, T]

Once the acquisition function is defined and the parameters set in the YAML configuration file, we can initiate the single-objective problem we are optimising for and the planner method.

optimizer = SequenceGA(algorithm='GA', period=15, design_protocol_filename='design_protocol.yaml')
planner = Planner(acq_fun, optimizer)

Now it is time to run the optimization!!

peptides = seed_library.copy()
pic50_scores = pic50_seed_library.copy()

# Here we are going to do 3 DMT cycles
for i in range(3):
    # Run optimization, recommend 96 new peptides based on existing data
    suggested_peptides, _ = ps.recommend(peptides, pic50_scores.reshape(-1, 1), batch_size=96)

    # Here you can add whatever methods you want to further filter out peptides
    
    # Get the pIC50 (Make/Test) of all the suggested peptides using the MHC emulator
    # WARNING: This is for benchmarking purpose only. This 
    # step is supposed to be an actual lab experiment.
    pic50_suggested_peptides = lpe.score(suggested_peptides)
    
    # Add all the new data
    peptides = np.concatenate([peptides, suggested_peptides])
    pic50_scores = np.concatenate((pic50_scores, pic50_suggested_peptides), axis=0)

Citation

If mobius is useful for your work please cite the following paper:

@misc{eberhardt2024combining,
  title={Combining Bayesian optimization with sequence-or structure-based strategies for optimization of protein-peptide binding},
  author={Eberhardt, Jerome and Lees, Aidan and Lill, Markus and Schwede, Torsten},
  url={https://doi.org/10.26434/chemrxiv-2023-b7l81-v2},
  doi={10.26434/chemrxiv-2023-b7l81-v2},
  year={2024}
}