Method to Classify Microbes in the Context of Probe Cross Hybridization Using Machine Learning Spectral Unmixing
This technology uses machine learning to accurately classify closely related microbes that have high homology in their 16S ribosomal RNA sequences.
Microbes have high sequence homology in their 16S ribosomal RNA when they are closely related – for instance, when they are members of the same strain or species. In addition, many 16S rRNA targeted oligonucleotide DNA microbes will cross-hybridize with closely related organisms, further compounding the challenges to accurate identification. This often makes these microbes difficult to distinguish and properly classify. Fluorescence in situ hybridization probes are generally sensitive to high sequence homology. As a result, researchers are actively looking for better tools for classifying organisms with cross-hybridizing probes in fluorescence microscope images of unknown mixtures of these microbes, including in natural communities.
This technology combines a bacterial fluorescence in situ hybridization probe design procedure, a sequential excitation imaging procedure, and machine learning to classify microbes that have extraordinarily high homology in their 16S ribosomal RNA sequences. Probes are designed to have 3 or more mismatches out of 18 nucleotides. Individual isolates of each species to be labeled are grown in the laboratory and fixed. Then, these cells are labeled with all probes to be used in the experiment, which intentionally induces cross-hybridization of highly sequence-similar probes. Images are then acquired with commercial microscopes using multiple separate excitations. The images are concatenated and machine learning is used to train a model to classify microbial species in new mixed images not previously seen by the model.
Photo for reference only, not a depiction of the invention.
The main advantage of this technology is its superior ability to distinguish and properly classify closely related microbes.
The primary application for this technology is the classification of closely related microbes that have high homology in their 16S ribosomal RNA sequences.
TRL 3 – Experimental proof of concept
This technology is available for licensing.
Patent Information:
| App Type |
Country |
Serial No. |
Patent No. |
Patent Status |
File Date |
Issued Date |
Expire Date |
|