Life sciences have benefitted immensely from advances in artificial intelligence. AI has a lot of potential to enhance and accelerate drug discovery — the process of identifying potential medicines. In January 2020, British start-up Exscientia and Japanese pharmaceutical firm Sumitomo Dainippon Pharma used AI to develop a drug for OCD. The typical drug development processes take around five years to reach the trial stage, but this drug took only a year.
Cheminformatics has grown by leaps and bounds in the last decade. Below, we have listed 6 AI-powered tools used for drug discovery.
AlphaFold
Proteins, made up of chains of amino acids, are the building blocks of life. What a protein does is largely a function of its unique 3D structure. In Critical Assessment of Structure Prediction (CASP), DeepMind’s AlphaFold has been recognised as a solution for the protein folding problem.
AlphaFold developed an attention-based neural network system to interpret the structure of protein’s spatial graph. It used evolutionarily related sequences, multiple sequence alignment (MSA), and a representation of amino acid residue pairs to refine this graph. The AI system developed strong predictions of the underlying physical structure of the protein through iterating the process.
DeepMind is looking into how protein structure predictions can help us learn more about diseases by identifying the proteins that fell into disrepair. Such insights could accelerate drug development efforts. Protein structure prediction is also helpful in pandemic response efforts.
Learn more here.
DeepChem
DeepChem is an open-source deep learning framework for drug discovery. The python-based frame-work offers a set of functionalities for applying deep learning in drug discovery.
It uses Google TensorFlow and scikit-learn to build neural networks for deep learning. It also makes use of the RDKit Python framework for basic operations on molecular data, such as converting SMILES strings into molecular graphs.
Learn more here.
ODDT
The Open Drug Discovery Toolkit is an open-source tool for computer aided drug discovery (CADD). ODDT uses machine learning scoring functions (RF-Score and NNScore) to develop CADD pipelines. It is provided as a Python library.
ODDT is built to support different formats by extending the use of Cinfony – a common API that unites molecular toolkits, such as RDKit and OpenBabel, and makes interacting with them more Python-like. All atom information collected from underlying toolkits are stored as Numpy arrays, which provide both speed and flexibility.
Open Drug Discovery Toolkit is released on a permissive 3-clause BSD license for both academic and industrial use. ODDT’s source code, additional examples and documentation are available on GitHub (https://github.com/oddt/oddt).
Learn more here.
Cyclica
Bio-tech company Cyclica’s MatchMaker harnesses reams of biochemical and structural data to assess candidate molecules against the entire proteome in quick time. POEM (Pareto-Optimal Embedded Modeling) is a parameter-free supervised learning approach to build property prediction models with more interpretability and less overfitting.
Naheed Kurji, the CEO of CyclicA said: “If you’re designing a molecule, it behooves you to consider the other 299 interactions that could have disastrous effects in humans.”
Image source: https://www.cyclicarx.com/science
Leveraging MatchMaker and POEM, Cyclica’s Ligand Design and Ligand Express platforms design novel, drug-like chemical matter by simultaneously prioritising compounds based on their on- and off-target polypharmacological profiles and their ADMET properties.
Image source: https://www.cyclicarx.com/science
Learn more here.
Exscientia
Exscientia is a pharmatech company leveraging AI to discover and design medicine in quick time. Exscientia’s AI platform has now designed two drugs that are in Phase 1 human clinical trials.
Exscientia has built AI systems to learn from data and apply the learning through design iterations.
Image Source: https://www.exscientia.ai/centaurai
Learn more here.
AMPL
The ATOM Modeling PipeLine (AMPL) is an open-source, modular, extensible software pipeline for building and sharing models to further in silico drug discovery.
AMPL extends the functionality of DeepChem and supports an array of machine learning and molecular featurization tools. It is an end-to-end data-driven modeling pipeline to generate machine learning models that can predict key safety and pharmacokinetic-relevant parameters. AMPL is benchmarked on a huge pool of pharmaceutical datasets and against a wide range of parameters.
Learn more here.
