AI in Drug Discovery & Development
AI in Drug Discovery & Development
Image Source: http://www.ihpasf.com/id/artificial-intelligence-making-its-way-into-pharmacy
Artificial Intelligence has numerous use cases when we talk about Drug Discovery and Development processes. Artificial Intelligence has been able to greatly advance the age-old pharmaceutical research methodologies and upgrade them from traditional approaches.
Drug Development is based on Molecule development. There are 2 approaches to develop molecules:-
1. Structure-based
This approach requires to have the knowledge of structure of both target and ligand
It deals with methods for free-energy binding calculation, protein-ligand docking, Dynamics of molecules etc.
2. Ligand-based
Ligand based approach uses ligand information to predict biological response.
What is a ligand?
It is an Ion/Molecule which binds with the central atom to form a complex coordination.
Drug Discovery Pipeline
Image Source: https://pubs.acs.org/doi/10.1021/acsmedchemlett.8b00437
Energy Evaluation between Molecules
It relies on the empirical scoring function/ force fields. During drug discovery and development, identifying the force fields, parametrizing and transferring them is a difficult task when there are a million combinations of possible ligands. Parametrizing and transfer of force fields is difficult as converting high dimensional quantum physics into an analytic functional form is quite a tedious task. A zoo of force fields are created to assist this task.
Challenging Task in Drug development
Understanding the dynamics of Force field, where and how it works is a big challenge
Force fields are very fast but they can perform poorly out of their environment/setup. So, it is difficult to get accurate force field measurement.
The computational costs of quantum mechanical methods required to be designed is high and time consuming.
These 2 cause major delays in drug development.
Artificial Intelligence improves the role of computational methods in the field of drug discovery and development. Artificial Intelligence and Big data together is termed as the 4th Industrial Revolution.
Following Machine Learning models are used to measure Quantum Mechanical Properties.
Kernel Ridge Regression
Neural Networks
Gaussian Process Regression
Advantages — Using the above models, results in high accuracy of quantitative property measurement and results in lowering numerical complexity.
Traditional Methods used in Drug Development
Ab initio -It is a quantum chemistry method which attempts to solve the electronic Schrodinger’s equation given the positions of the nuclei and the number of electrons in order to yield useful information such as electron densities, energies and other properties of the system. The ability to run these calculations has enabled theoretical chemists to solve a range of problems
DFT — Density Functional Theory is a computational quantum mechanical modelling method used in physics, chemistry and materials science to investigate the electronic structure (or nuclear structure) (principally the ground state) of many-body systems, in particular atoms, molecules, and the condensed phases
The above mentioned Machine Learning models perform better than ab initio and DFT. The Machine Learning models are able to better determine complex relationships among data better than physical on strained approximation methods like Force-fields and Semiempirical QM.
A disadvantage of Machine Learning model is that they heavily depend on the quality and quantity of training data. A vast amount of reference data is required to train general purpose models.
ANI-1 is the first example of Universal extensible Neural Network Molecular potential. The training data for ANI-1 was about 22 Million small molecule conformations.
Output of ANI-1 — The model is able to predict energies on larger systems and new test data with an applicability on molecules upto 70 atoms.
Artificial Intelligence for Primary Drug Screening
Computer Vision models — Used for the identification of distinct objects/features in recognizing images. Consider Breast Cancer Detection. Following are the steps involved:-
Image Processing -Cell Images are separated from the background, advanced image processing techniques to improve the resolution of images.
Extraction of Tamura (Tamura features are important texture features which are based on human visual perception) and Wavelet texture feature (Wavelet transforms [6]represent an image in a space whose coordinate system has an interpretation that is closely related to the characteristics of a texture)
Principal Component Analysis — To reduce the dimensions of extracted image features
Classify cell types using Least Squares Support Vector Machines
Artificial Intelligence in Pharmaceutical Devices
Image Activated Cell Sorting Devices (IACSD) -These are high speed digital image processing and decision making devices which use Convolutional Deep Neural Network as its architecture.
AI is also used for the interpretation of Electrocardiography (ECG) of clinical diagnosis. It has helped increase the accuracy and scalability of automated ECG analysis.
Artificial Intelligence in Secondary Drug Screening
AI is used in the predicting physical properties in the drug development/ design phase. The most probable drug candidates can be identified based on the following features:-
Bio-availability
Bio-activity
Toxicity
Melting Point (Ease of Dissolving) and Partition Coefficient (Relative solubility between water and oil) — greatly influence Bio-availability. These 2 properties are important features considered in the design of new drugs.
Machine Learning models uses the following properties to assist drug screening processes:-
Molecular fingerprint — [simplified molecular input line entry System (SMILES)] string
Potential Energy Measurements
Molecular Graphs
Coulomb Matrices
Happy Reading :)
References
Transforming Computational Drug Discovery with Machine Learning and AI
In this Viewpoint, we discuss the current progress in applications of machine learning (ML) and artificial intelligence…pubs.acs.org
IHPASF | Artificial Intelligence: Making its way into Pharmacy
Artificial intelligence is a branch of computer science that deals with problem-solving applications by the aid of…www.ihpasf.com