The process of drug development is a fundamental pillar of modern healthcare. But the laborious journey through complex mazes of research, development, and testing  takes several years. Enter Artificial Intelligence (AI). In recent years, it has enriched industries and boosted human potential, more than ever before. How can AI reshape the pharmaceutical industry? To answer this, let us first understand typical steps involved in creating new drugs.

The Traditional Route of Drug Development

Traditionally, drug development follows a linear trajectory, composed of various phases.

Target identification and validation: The process begins with identifying a specific biological molecule, called the "target". This could be a specific protein, enzyme, receptor, or a molecular component that plays a key role in a disease's development or progression. Researchers study the target to understand its structure and function. After confirming that modifying the target can achieve the desired therapeutic effect, it can be used to create drug interventions.

Lead molecule discovery and optimisation: With the target pinpointed, researchers look for molecules that have biological activity against it. This is done by screening for molecules that interact with the target, known as "leads". 

Two main classes of compounds are considered: small molecules and biologics. Small molecules have been used as drugs for over 200 years. These are often organic molecules which can be chemically synthesised. Some examples of small molecule drugs are aspirin, penicillin and paracetamol. Biologics are a relatively newer class of therapy. They are larger, more complex molecules often derived from living sources or engineered using biological systems. They target proteins, receptors, or cells in a more specific manner. Biologics include monoclonal antibodies, hormones, vaccines, cell and gene therapies. 

The choice between small molecules and biologics depends on the target, disease, and desired mode of action. After selecting a lead, scientists optimise certain attributes of the molecules. This includes their potency, how specific they are to the target, and their safety profile. It is essential to strike a balance between improving the molecules's therapeutic properties while avoiding undesirable side effects.

Preclinical Testing: The optimised lead molecules are tested using cell cultures and animal models. These studies provide data on the drug's pharmacology and toxicology. For successful leads, the process of formulation, purification and manufacturing is streamlined.

IND Application: After preclinical testing, the pharmaceutical company submits an Investigational New Drug (IND) application to the office of the Drugs Controller General (DCG), India or Food and Drug Administration (FDA) in the US. This application contains a comprehensive package of data, including preclinical study results, manufacturing details, and a clinical trial plan. This information is used to decide whether the drug candidate should go to the clinical trials. 

Clinical Trials: If IND is approved, the safety and efficacy of the drug is tested in human subjects during clinical trials. These studies help to determine if a new drug is safe and effective, before being approved for public use. There are 3 main phases of clinical trials – phases 1 to 3. Each clinical trial is uniquely designed, considering the drug, its properties and effects. Clinical trials are done in collaboration with medical institutions, researchers, regulatory authorities, and sometimes drug companies. In India, the Central Drugs Standard Control Organization (CDSCO) regulates clinical trials.

Regulatory Approval: Regulatory bodies review all the drug related data to see if the benefits outweigh its risks for the intended use. This involves careful analysis of clinical trials, manufacturing practices, and proposed labelling. After the regulatory authorities are satisfied, they grant approval for the drug to be marketed and sold. 

The Rise of The AI

From entertainment to education, artificial intelligence (AI) and machine learning (ML) have integrated into our lives- opening unprecedented possibilities. In the last few years, the transformative power of AI-ML has been remarkable. Its ability to comb through vast datasets, recognise patterns and predict outcomes has completely changed how we solve problems. Big data is the new currency, and the pharma giants are leveraging this by incorporating ML in various aspects of their operations. Meanwhile, new startups are joining the pharmaceutical industry with fresh ideas and approaches. AI-driven R&D deals in drug discovery were at a record-high in 2022, with $13 billion in deals. 

Pharmaceutical companies and startups are adopting AI in all stages of drug development. 

Target identification, optimisation and validation: AI can identify novel targets from biological data. ML models can be used  to identify specific regions on the target protein's surface. These patches can then be replicated to make the target more suitable for interaction with drugs. AI can also be used to analyse genetic, proteomic and clinical data to validate potential targets, improving target selection. 

Lead molecule discovery and optimisation: AI can make informed decisions about selecting and designing candidate leads for both small molecule and biologic drugs. 

High-throughput screening: ML algorithms can predict the biological activity and likelihood of a drug binding to target proteins. This helps with virtually screening large libraries of small molecules to find hits. Therefore, reducing the number of compounds to be experimentally tested, saving time and resources.

Novel small molecule discovery: AI can learn patterns and relationships within molecular structures from massive datasets. This enables generation of novel compounds with the potential to become small molecule drugs. AI-driven approach significantly expands the exploration of chemical space, uncovering unique structures that might not be evident through traditional methods

Generics: Generics are small molecules, designed to be identical, or bioequivalent, to drugs that are already on the market. AI predicts how different components interact in generic drugs and ensures they are similar to the reference product.

Lead optimisation: ML-driven models predict how chemical modifications impact a small molecule's properties. This accelerates lead optimisation by suggesting modifications that enhance efficacy and safety.

Biosimilars: Biosimilars are biologics that are highly similar to an already approved reference biologic drug. AI compares biosimilars to reference biologics to ensure structural similarity and predict the functional properties of biosimilars. 

Biobetters: Biobetters are altered versions of current biologic medications (including reference drugs and biosimilars). AI uses molecular and disease data to find targets for biobetters. ML-guided protein engineering helps in designing biosimilar and biobetters structures for enhanced efficacy and reduced side effects.

Novel biologics discovery and optimisation: AI uses predictive modelling and structural analysis to pinpoint areas of improvement in biologic molecules. It then proposes modifications while preserving the molecule's desired function and therapeutic properties. This approach helps design novel, more effective biologics. 

Repurposing existing drugs: ML is used to analyse databases to identify approved drugs that could be repurposed for new indications, saving time and costs compared to developing entirely new drugs, in both small molecules and biologics.

Drug formulation: AI-assisted algorithms help refine formulations that can improve solubility, stability and bioavailability.

Preclinical testing: ML models analyse diverse datasets to predict toxicity and potential side effects of drug candidates, enhancing safety assessments and expediting preclinical research.

Clinical trials: AI optimises trial design by examining patient information, predicting patient responses, and contributing to more efficient and effective trials. This increases the likelihood of successful trials and reduces trial duration.

Personalised medicine: AI tailors treatments to individual patients based on genetic, molecular, and clinical data, improving therapeutic outcomes. ML also predicts drug-drug interactions, reducing the risk of adverse effects and enabling personalised treatment plans.

Novel Antibody Discovery 

Antibody therapy, also called immunotherapy or biologic therapy, involves using natural or laboratory-made antibodies to target and treat diseases. Because of their high specificity, antibodies are on their way to becoming the largest component of the biologics landscape. 

However, the time and money required in the conventional process of antibody discovery renders its potential to revolutionise healthcare untapped. The classical method of discovering antibodies is injecting an animal with the chosen target antigen to harvest immune cells that produce antibodies. This process is not only lengthy and laborious, but also suffers from a low success rate. To close the gap in the vertical of the rapidly growing market of novel antibody design, many companies are using AI-driven approaches. This method has made discoveries faster and efficient, while increasing the likelihood of success.

A common strategy is to predict how mutations or modifications to an existing antibody might impact their function and properties using ML models. AI suggests changes that enhance its attributes while preserving its therapeutic function. Most antibody companies today are using AI to optimise biologically derived or synthetically generated antibody sequences against known target antigens. However, these approaches do not come without caveats. 

• After synthetic design and AI-predicted optimisation, there is an immense load on experimental validation- a resource-intensive phase. This is where the functional and therapeutic properties of the designed molecule are tested. 

• Often, these synthetic molecules need to be significantly modified to achieve desired properties, for instance, a good degree of binding to the antigen and minimal toxicity. 

• Sometimes, the molecule loses its therapeutic properties in this process and cannot be developed into a viable drug.  

• And lastly, there is still a heavy reliance on biological sources for discovering novel antibodies.

To eliminate the reliance on animals and biological sources and increase efficiency, the solution is to use the power of AI to generate novel antibody drugs from scratch, designed to bind the target molecule specifically. This will increase the efficiency of the drug discovery and development process, thereby reducing both time and cost. 

immunitoAI’s proprietary AI-platform is working towards this goal. For a deeper dive into immunitoAI's role in antibody discovery, check out our related blog.

Related Blogs