Artificial Intelligence (AI) in drug discovery

Discover how Artificial Intelligence is transforming and speeding up drug discovery
Tue 15 Dec 2020

The discovery and development process for a new drug can be extremely long, expensive, and challenging. The process takes an average of 10-15 years, with only 10% average success rate. Due to the nature of the drug development process, which is long and with a low success rate, the average cost for bringing a single drug on the market has reached $2.8 billion - with the early drug discovery phase being the most expensive stage of the process (33% of the total development costs).

Based on a report by Deloitte, AI-based drug discovery is the most promising solution for tackling early drug discovery and development. It is projected that the market size of AI-based drug discovery will reach up to $1.43 billion by 2024, with an annual increase of 40.8%.

Now, let’s take a look at how artificial intelligence can enhance the drug discovery process, some potential use cases, and the benefits of using AI for drug discovery.

How is Artificial Intelligence accelerating drug discovery

Artificial Intelligence has the potential to transform the drug discovery process. It can make it more efficient and more effective, thus benefiting all parties involved - from companies developing new drugs to patients in desperate need of viable treatments.

The health and biotech sector is becoming increasingly rich in data, and these large datasets require advanced analytical methods that can help understand the relationships within the data. However, the lack of advanced technologies limits the drug discovery process, making it a time-consuming and expensive task. Artificial intelligence and machine learning can help address these limits by processing biomedical data to spot patterns of biochemical properties, identify novel interactions of different compounds, and make predictions. For example, AI can extract meaningful information from large datasets, e.g. a dataset of RNA sequencing can be used to identify genes whose expression correlates with a given cellular condition.

AI can be used in many stages of the drug discovery process. Some examples include:

  • AI virtual high-throughput screening,

  • Predicting target protein structure with AI,

  • Predicting bioactivity with AI,

  • Toxicity prediction with AI, and more.

Let’s take a look at them in more detail below.

AI virtual high-throughput screening

High-throughput screening (HTS) is a drug discovery process that enables automated testing of a large number of chemicals and/or biological compounds for a specific biological target. The primary goal of HTS is to identify potential candidates that affect the target in the desired way, (so-called ‘hits’ or ‘leads’) through compound library screening. This is usually achieved through the use of robotics, liquid handling devices, plate readers as detectors, software for instrumentation control, etc. To run effectively, high-throughput screening requires highly specialized facilities, it can be an expensive process, and not always showing the most accurate results.

Artificial intelligence virtual high-throughput screening can increase both the performance and decrease the cost of finding new drug ‘hits’, in the following way:

  • Increase the accuracy and the speed of the screening process,

  • Enable screening of large data libraries,

  • Enable screening of expensive bioassays, bioassays requiring dangerous reagents, and compounds that are complex to synthesize.

Kantify has developed its own state-of-the-art solution for drug discovery. Discover more about virtual high-throughput screening with AI in one of our use cases.

Predicting target protein structure with AI

Currently, there are around 200 million known proteins, with another 30 million found every year. Each one has a unique 3D shape that determines how the protein works and what it does. While developing a drug molecule, it is essential to assign the correct target for a successful treatment. Numerous proteins can be involved in the development of a particular disease, and in some cases, they can be overexpressed. Hence, for successful selective targeting of a disease it is important to predict the structure of the target protein, in order to design the appropriate drug molecule.

AI can assist in structure-based drug discovery by predicting the 3D protein structure, and later predicting the effect of a particular compound on the target protein. For example, DeepMind’s AlphaFold AI-based algorithm can predict the 3D target protein structure by analyzing the distance between the adjacent amino acids and the corresponding angles of the peptide bonds.

Predicting bioactivity with AI

The efficacy of drug molecules depends on their affinity to the target protein or receptor. Drug molecules that do not show any interaction or affinity towards the target protein will not be able to deliver the expected therapeutic response. Hence, it is vital for a drug molecule to have drug target binding affinity (DTBA) with the target protein/receptor, in order to create new and successful drugs. Artificial Intelligence can help measure and predict the binding affinity of a potential drug by looking into the features or the similarities of the drug and its target. For example, AI can help predict the binding affinity by exploiting the geometric binding site properties and non-covalent interaction patterns. Drug features from SMILES and extended connectivity fingerprint can also be considered to calculate and predict how a certain molecule will interact with the wanted target.

Toxicity prediction with AI

The prediction of the toxicity of any potential drug molecule is vital to avoid toxic effects and is essential for whether the drug molecule will continue to the next stage of the drug development process. In common practice, the toxicity is usually observed either in in vitro studies using cells/cell lines, or in in vivo exposure in trial experiments - making it a long and expensive process. AI-based algorithms can look for similarities among compounds, and predict the toxicity of the compound based on input features. For example, a machine learning algorithm named DeepTox was built as an initiative from the US Food and Drug Administration (FDA), the Environmental Protection Agency (EPA), and the National Institutes of Health. The algorithm was used to forecast the toxicity of 12 707 environmental compounds and drugs, and could efficiently predict the toxicity of a molecule - outperforming all traditional methods of computing - by looking at 2500 predefined toxicophore features.

Benefits of applying AI to drug discovery

The use of AI in drug discovery has the potential to revolutionize the current time and scope of drug discovery. Particularly AI can help:

  • Increase precision: AI has a higher predictive power to define meaningful interactions in a drug screen compared to the traditional drug discovery methods. For example, it can be used to identify novel drug structures/compounds, and help improve the ability of that novel compound to engage its target. Ultimately, AI can help increase precision, reduce false positives, and improve the design of the parameters of the assay in question.

  • ** Find new drugs cheaper and faster**: Because of the advancements in processing power, AI-based algorithms can make drug discovery a much cheaper and much faster process. For example, AI can screen a compound library of several billion molecules, and significantly reduce the time needed for in vitro efforts and for preclinical candidates to be identified (compared to conventional methods).

  • Empower virtual labs: It is not uncommon for companies to not have in-house laboratory plants, equipment, and instruments needed to run experiments. Fortunately, AI has the potential to move drug screening to a virtual lab, where the results of a screen can be obtained with greater speed, and without the need for extensive costs and long manpower hours.


Are you interested in finding out more on how you can use Artificial Intelligence for drug discovery? We have developed an expertise in helping biotech companies leverage AI to bring new drugs to the market, faster, cheaper, and more efficiently.

Let’s get in touch to discuss your challenge in more detail!