Robots reproducing results a game changer for life sciences
Japanese researchers automate procedures to boost accuracy and reliability
TOKYO -- A group of researchers in Japan has managed to automate key procedures for life science experiments, raising expectations of higher accuracy and reliability.
The group, led by researchers at the National Institute of Advanced Industrial Science and Technology (AIST), have developed a program for experiments that, due to human factors, are often very difficult to reproduce.
Using robots, the same experiments were conducted multiple times producing almost identical results.
Automation of procedures that are susceptible to human error is expected to help improve efficiency in the development of new drugs, as well as prevent research fraud.
Food company Ajinomoto and research institution Riken also took part in the project with AIST, alongside Keio University, Kyushu University and Tokyo Medical and Dental University.
The researchers looked at combinations of procedures, such as like cultivating and transferring cells, as well as treating samples in a centrifuge, and quantified the movements involved.
Each participating organization installed the program in commercially available dual-arm robots and conducted the same experiments independently in their own labs.
In an experiment on amplifying a specific gene, for example, the tests conducted at all the labs resulted in the amplification of the same gene. In a test to select compounds that bind with cancer cells from 48 different types, all the robots selected the same three.
These results show that robots controlled by the program can reproduce results from experiments.
In an experiment to measure the amounts of various proteins in a cell, the measurements of proteins with high volumes varied by only a few percent.
Done by hand, it is extremely difficult to measure proteins found in small volumes. But the robots, however, could measure the amounts of such proteins by discrepancies of about 20%.
One distinct advantage of the program is that it can be installed into the various machines researchers use and requires no knowledge of robotics.
The system is applicable to some 80% of life science experiments, according to Toru Natsume, director of the Molecular Profiling Research Center for Drug Discovery. The remaining 20% are infrequent procedures, for which individual institutions would need to create the necessary programs independently.
The robots used in the project are sold for about 100 million yen ($890,000) each by Robotic Biology Institute, a startup linked to AIST. The company has also created a system that allows drug makers to test its robots in their labs.
"Using five or so of the robots can generate great synergistic effects," said Natsume.
Subtle differences that would seem insignificant to many, such as the way a researcher shakes a container, can lead to discrepancies in results. One report says more than 60% of experiments described in research papers in this field cannot be reproduced. The problem has been hampering efforts to develop new medical treatments and leads to the proliferation of falsified results.
In response to the problem, the U.S. National Institutes of Health has embarked on developing guidelines for life science experiments.
Robots, however, completely take out the element of human error.
If researchers publish numerical data about procedures involved in their experiments, others can use robots run by the program to confirm the results.