Infographic: A Chart That Helps Predict Drug Side Effects

Prescription drugs are still in the dark ages, but as we wait for science to catch up, smart modeling predicts problems before they happen.

There’s nothing more depressing or comical than drug commercials. After promising us a better life--vacations by pools, hugs from spouses, and plenty of golden retrievers--we’re hit with a long list of nauseating side effects. Blurry vision. Stomach bleeding. Erections that last days. We begin to wonder, what the heck are they trying to sell us?

Click to enlarge.

That’s a question that a team of scientists from the UCSF School of Pharmacy, Novartis Institutes for Biomedical Research, and SeaChange Pharmaceuticals is trying to answer. Since most drugs are more like shotguns than sniper rifles, hitting the intended target but possibly several others in the process, they developed a model to track and predict the side effects of drugs.

It’s based upon UCSF’s “similarity ensemble approach” (SEA), which compares drug shapes to assess similarities. We know that some drugs definitely interact with some proteins, but that knowledge is limited, so SEA piles all the drugs and their known side effects into one big pool, along with all their known side effects. The result is this incredible infographic. It’s a map of 1,241 possible side-effect targets (the innocent bystanders) for 656 drugs on the market today.

The gold circles are the drugs. The molecular targets are blue. The gray lines signify that we know a drug hits that target. The red line is the fun part--that’s the prediction of where drugs may hit, based upon their similarity.

“In a sense, we’re rediscovering pharmacology, and augmenting it with computational statistics,” explains SeaChange’s Michael Keiser. “Before this newfangled era of molecular biology, pharmacologists had to guess at and even define therapeutic targets solely by drug effects … by automating and quantifying what was once a manual, pharmacological process in these models, we can get a head start on a computational system to relearn and then expand on what only expert pharmacologists once knew.”

Basically, whereas 3-D computer modeling can eventually unravel all of these complex drug interactions at the molecular level, SEA can make an extremely educated guess with the information we have today. And in the team’s recently published paper, they revealed the discovery of 151 new drug side effects that were later confirmed by lab testing.

One such example was regarding chlorotrianisene (CTX). CTX is a synthetic estrogen used in hormone replacement therapy and some cancer treatments, but it was attributed to abdominal pain (sometimes severe). SEA modeling discovered that CTX was binding to a totally unintended target, the same enzyme that aspirin inhibits (also known for abdominal pain) … with 10x the strength it binds with the intended estrogen receptor.

“We tested CTX in freshly donated human blood and found it was indeed acting like aspirin, and very strongly,” writes Keiser. “So now we have an idea what the reason is for this serious side effect--which has caused hospital visits--and perhaps how to avoid it. And that’s just one red line, so to speak.”

[Hat tip: Gizmag]

Add New Comment