Nanopore biodetection project wins Proof-of-Concept award

9/9/2016 Maeve Reilly, Beckman Institute

"Machine Learning for Nanopore Biodetection" has direct applications in DNA sequence-dependent diagnosis and in detection and treatment of certain diseases.

Written by Maeve Reilly, Beckman Institute

ECE ILLINOIS Professor Jean-Pierre Leburton and Assistant Professor Lav R Varshney recently received an Illinois Proof-of-Concept Award from the Office of Technology Management for a project entitled “Machine Learning for Nanopore Biodetection.” The Illinois Proof-of-Concept programs fund development projects that demonstrate an innovation's market viability to potential investors and partners. Projects consist of a defined set of milestones that, when completed, help overcome a specific hurdle to an innovation's transfer outside of the University.

Jean-Pierre Leburton
Jean-Pierre Leburton
Leburton, a Beckman researcher and member of the Nanoelectronics and Nanomaterials Group, has been looking at DNA sequencing using nanopore technology, which promises to deliver low-cost, fast, reliable and highly accurate sequencing of a person’s whole genome. The basic design is composed of a solid-state, multilayer semiconductor membrane that uses nanopores about the diameter of a single DNA molecule (about a billionth of a meter wide). Employing electric fields, single DNA molecules are passed through the nanopore and a detector reads the sequence.

While the technology promises to provide a revolution in personalized medicine, it is not without its challenges.

“When the DNA goes through the nanopore there is a lot of noise around it,” said Leburton. “This is due to jittering or flossing caused by the thermal motion of the DNA in water.” That noise affects the accuracy of the detection processes of the DNA molecule and the identification of any abnormalities in the molecule.

The researchers looked specifically at the methylation of DNA, the process in which gene transcription is repressed and could result in diseases such as cancer.

“We can attach a protein to the methyl group, and we can actually detect the protein because it is a big object,” explained Leburton. “The measurement is still affected by noise, but, because the protein is large, we can actually recognize it.” However, identifying smaller bodies that can alter DNA is still a challenge

Lav R Varshney
Lav R Varshney
Varshney, a member of Beckman’s Image Formation and Processing Group, told Leburton that he could design a simple algorithm that could help filter the noise, allowing for the identification of any alterations that might be present in the DNA.

“The basic idea is to try to detect the sequence of methylated bases using statistical signal processing and machine learning techniques, a little bit like detecting targets in radar signals,” explained Varshney.

The researchers believe that the new technology has direct applications in DNA sequence-dependent diagnosis and in detection and treatment of certain diseases such as cancer.

 


Share this story

This story was published September 9, 2016.