Five reasons why Cognitive Modeling is the best technology for DNA Sequencing
The promise of Nanopore technology is to sequence long strings of CCDNA -up to 50,000 bases-. Current DNA sequencing technology is able to process strings of up to 100 bases, and partial genome sequencing can take months to do. Northshore Bio – a Seattle based startup- is in a very unique position in the industry, with the beginnings of a stable and inexpensive CMOS chip with an array of Nanopores for molecule identification.
Qualitative Artificial Intelligence ( QAI) is proud to announce the successful development of DNA sequencing software using Cognitive Modeling for one of our clients - Seattle based, NorthShore Bio
The challenge: Although very promising, Nanopore technology is still in its infancy and the signal obtained from the nanopore was different from one run to the next. As each molecule passes through the nanopore, a signal peaks is created. It is difficult to identify by eye or quantitative methods the peak characteristics that corresponds to each DNA base. At this point in development, there were too many unknowns to apply traditional machine learning techniques.
The solution: Cognitive Modeling is the perfect fit for this time series challenge. The shape of each peak is qualitatively described. Qualitative modeling is independent of the variations in amplitude and distance of the signal. Peaks with common qualitative characteristics are combined into groups called classes, and archetype peaks are automatically created. Reports are generated to analyze the entire process.
Time: The development of the project from concept to end product took around 6 months, as it was scheduled, even though it was more of a research project, with many unknown aspects.
Five reasons why Cognitive Modeling is the best technology for Nanopore-based DNA sequencing:
- Cognitive Modeling is a formalization of the intuitive way humans classify the peaks - automatizing the process that NorthShore Bio was doing manually.
- Cognitive Modeling does not require big amounts of data to learn.
- Cognitive Modeling does not require long periods of time to learn.
- Cognitive Modeling is transparent – people can see and analyze the decision processes which lead to peak classification (on the contrary to other techniques of machine learning which act as black boxes).
- Cognitive Modeling solutions are elegant, simple, and fast - making the promise of Nanopore-base DNA sequencing a reality now.
Watch for this new disrupting technology changing the way we do DNA sequencing in the near future!
If your company is facing challenges that might be resolved with Cognitive Modeling, do not hesitate to contact us at [email protected].
Doctor in Cognitive Science working in Artificial Intelligence Engineering
8 年May I ask what part of the modeling you mention in the editorial above is 'cognitive'? Computational cognitive modeling is already a common term used to describe the practice of using cognitive architectures (such as ACT-R, Soar, EPIC, CLARION, etc.) in order to build models of cognition using experimental data. I humbly do not see what the relationship with cognition is in the project mentioned above. Perhaps you refer to so-called "deep learning", the cognitive component of which is the reliance on biologically-inspired algorithms such as artificial neural networks?