Therapeutic Discovery

Predicting Therapeutic Antibody Properties with Machine Learning

Once trained, machine learning algorithms can predict the properties of new antibody candidates, greatly speeding up the drug discovery process.


Machine learning algorithms can analyze large volumes of data, identify patterns, and make predictions based on those patterns. In the context of therapeutic antibodies, these algorithms can be trained on databases of existing antibodies and their properties. They then learn relationships between the structure of these antibodies and their effectiveness against various diseases. Once trained, they can predict the properties of new antibody candidates, greatly speeding up the drug discovery process.

 

Examples of Machine Learning Algorithms in Action

  1. Deep Learning for Antigen Specificity: A study published in Nature Biomedical Engineering demonstrated the use of deep learning, a type of machine learning, to predict antigen specificity from antibody sequences. The researchers used a large dataset of antibody variants and trained a deep learning model to predict the specific antigens that each antibody would bind to. This approach could significantly streamline the process of identifying suitable antibodies for specific diseases.
  2. Predicting Antibody Aggregation and Viscosity: Another crucial aspect of therapeutic antibody development is predicting properties like aggregation and viscosity. An article in mAbs describes a machine learning model that was trained on a dataset of therapeutic antibody aggregation and viscosity data. The model was able to accurately predict these properties for new antibody candidates, which could help in the formulation development of protein therapeutics.
  3. Bioavailability Prediction Using Machine Learning: Ensuring adequate bioavailability, or the extent to which a drug reaches its intended site of action, is also a key factor in therapeutic antibody development. A study in Pharmaceutical Research demonstrated the use of machine learning to predict the subcutaneous bioavailability of monoclonal antibodies based on their properties. This could help in the early stages of antibody development by identifying candidates likely to have sufficient bioavailability.

Similar posts

StackWave Affinity™

Whether you're a CRO making customer deliveries, a start-up advancing towards IND, or an established biopharma supporting multiple programs at once, Affinity supports every team, at every step, on a single platform, for a single price.

Download the StackWave Affinity LIMS presentation

Sign up for notifications