Lawrence Livermore National Laboratory (LLNL) researchers say they’ve discovered a way to use supercomputers to boost drug safety by identifying potentially adverse side effects early on in the drug development process.
Medications alleviate pain and suffering, but the drug creation process misses side effects that kill at least 100,000 people a year, according to the journal Nature.
The LLNL researchers have developed a method that uses high-performance computing (HPC) systems to process proteins and drug compounds in an algorithm to determine which drug candidates can cause unanticipated side effects, according to a recent LLNL press release.
The drug discovery process starts by identifying the proteins that are associated with specific diseases. Candidate drug compounds are then combined with target proteins in a process called binding to determine the drug’s effectiveness or toxicity, the release said. For the pharmaceutical to work, the drug compounds have to bind with proteins.
This method can identify side effects with many target proteins, but there are unknown “off-target” proteins that may bind to the drug candidate and cause unanticipated side effects, the release said.
Pharmaceutical companies usually test only a small set of off-target proteins during the early stages of drug discovery because it’s too cost-prohibitive to test against a larger set of proteins. This can result in drugs reaching the marketplace with the adverse side effects going undetected during early testing and clinical trials, the release said.
To tackle the issue, LLNL researcher Monte LaBute and his team used supercomputers to examine 4,020 off-target proteins from protein databases, such as DrugBank, and indexed them to find 409 off-target proteins that have high-quality 3D crystallographic X-Ray diffraction structures that are essential for computational analysis, the release said.
The research team used Livermore HPC software to analyze the 409 off-target proteins with 906 FDA-approved drug compounds. Most of the calculations were run on LLNL’s Intel-based Sierra and Cab HPC systems.
“We then performed docking calculations on our supercomputers using the 3D structures for both drugs and proteins, trying all combinations of drugs against all of the proteins,” LaBute said in an email. “The output for these calculations was a numerical score that indicated how strongly a given drug bound to a particular protein.”
Separately, for these same drugs, the researchers extracted known drug-side effect associations from data found in the SIDER drug database, he said in an email. Finally, they fed the drug-protein docking score data and the drug side effect associations into machine-learning algorithms to determine which proteins were associated with certain side effects.
“This information is then used to construct models that can more accurately identify off-target proteins that cause side effects,” he said in an email.
The process provides pharmaceutical companies a cost-effective and reliable way to find off-target proteins that cause side effects, Labute said. “This approach using HPC and molecular docking to find ADRs (adverse drug reactions) never really existed before,” he said in the press release.
The researchers, who recently published their findings in the journal PLOS ONE, now want to expand their research to include more off-target proteins for testing and eventually screen every protein in the body, a process that might take 10 years.
“If we can do that, the drugs of tomorrow will have less side effects that can potentially lead to fatalities,” he said in a statement.
The post Deadly Drug Side Effects Targeted by Supercomputers appeared first on Go Parallel.