I’ve written before about some fascinating projects using machine learning to test the potential toxicity of medicines.  A good example of what’s possible is provided by a recent study published in Cell Chemical Biology.  The study reveals a big data based approach to detecting toxic side effects that would prohibit a drug from being used on humans before it gets to the expensive clinical trial stage.

Or you’ve got the recent Stanford University study, which highlights the potential of AI, even with relatively small amounts of data to play with.  The team used a new kind of deep learning known as one-shot learning that can do its stuff with relatively few data points.

“We’re trying to use machine learning, especially deep learning, for the early stage of drug design,” the team say. “The issue is, once you have thousands of examples in drug design, you probably already have a successful drug.”

Detecting toxicity

A recent study continues the advance in this field.  The researchers developed a novel approach to evaluating drug safety that detects stress on cells at a much earlier stage than existing methods that usually rely on detecting the death of cells.

The method utilizes fluorescent sensors that are turned on in the cell when misfolded proteins grow in number, which is an early sign of cellular stress.

“Drugs can cause proteins—which are long strings of amino acids that need to be precisely folded to function properly—to misfold and clump together into aggregates that can eventually kill the cell. We set out to develop a system that can detect these aggregates at very early stages and that also uses technology that is affordable and accessible to many laboratories,” the authors explain.

The approach is believed to be the first of its kind.  The researchers developed an unstable protein, called AgHalo, to test their method.  It’s designed with a fluorescent dye that becomes active in a water-repellent environment.  Such environments are usually buried deep within the structure of a protein that is properly folded.  When the protein begins to misfold, the dye interacts with the water in the protein and glows.

Put to the test

The system was tested out on five commonly-used anti-cancer drugs.  Whilst in previous drug safety tests none of the drugs reported high levels of cell death, the AgHalo sensor detected a noticeable level of protein stress.

“Because we tested the anti-cancer drugs at much higher doses than typically used for treatment, our results do not necessarily call into question the continued use of these drugs,” the authors say. “However, because protein stress from long-term treatments could have lasting effects, evaluating drugs with our new sensor will help in the development of safer drugs.”

The team believe their method can put a quantifiable figure on the protein stress within a cell, and can do so at a much earlier stage than existing methods.  This will allow researchers to study the various mechanisms that cause such stress, whether its’ bacterial infections, cancers or toxins, and develop suitable compounds to enhance the cell’s ability to handle such stresses.