The algorithm created through intensive learning finds potential therapeutic targets across the gene

credit : medium.com

A team of researchers from the New Jersey Institute of Technology (NJIT) and the Children's Hospital of Philadelphia (CHOP) have developed an algorithm through machine learning that can help predict the sites of DNA methylation - a process that can completely alter the structure of DNA. Missing disease-spreading patterns can be detected through traditional screening methods.

The journal Nature Machine Intelligence published the paper online this week.

DNA methylation is involved in many key cellular processes and is an important component in genetic expression. Similarly, defects in methylation‌ are associated with a variety of human diseases. Although genetic sequencing tools are effective in pinning the polymorphisms that cause a disease, the same methods fail to capture the methylation effect because individual genes still appear identical. In particular, considerable effort has been made to study DNA methylation on N6-adenine (6mA) in eukaryotic cells, including human cells, although the role of methylation in these cases remains unclear when genetic data are available.

Also Read : The biggest AI of 2035 is already here

"Previously, methods developed to identify these methylation sites in a gene were very conventional and could only see certain nucleotide lengths at a given time, so a large number of methylation sites were missing," Hecon said. Hakonerson, MD, Ph.D. D., Director of the Center for Applied Genomics (CAG) at CHOP and one of the senior co-authors of the study. "There is a need to develop a better way to identify and evaluate methylation sites with a tool that can detect these elements throughout the gene, which can have a strong functional effect and cause disease."

To address this problem plaguing the research community, CAI and its partners at NIIT have turned to intensive education. Zee Wee, a computer science professor at NJIT and senior co-author of the study, developed an intensive learning algorithm with Hakonorson and his team to track where these methylation sites originated, with the help of researchers using some nearby genes.

Wee called his software Deep6 MIR. Wee led the development of a neural network, a machine learning model that seeks to learn in ways similar to the brain, to track where these methylation sites are found. Neural networks have been used in cellular research, but this is the first application of the study of natural multicellular organisms.

Wee cited four advantages of the new method: the automation of a range of detail feature feature; Integrating a broad spectrum of methylation sequences into aving genes of interest; Initiating a view of latent sequence elements for commentary; And facilitates model development and assessment in large-scale genetic data.

The study team applied this algorithm to three types of representative organisms: A. thaliana, d. Melanogaster, and the first two eukaryotic E. coli. DP6MEA was able to detect 6mA methylation sites for solution of a single nucleotide or basic unit of DNA. Even in this preliminary diagnostic study, the researchers were able to visualize control patterns that they could not observe using existing methods.

“One limitation is that our proposed estimate is based entirely on regular information,” Wee said in a discussion statement for his study. "Whether a candidate is a 6mA site or not also depends on many other factors, including 6mA. Methylation is a dynamic process that varies with the cellular context. In the future, we may want to consider other factors [e.g. as genetic expression. Estimating 6mA in the cellular context by integrating other data We hope you like it. "

• As humans who have read a major area of our brain we can ‘recycle’
• The algorithm created through intensive learning finds potential therapeutic targets across the gene
• 5 Free Resources Every Data Scientist Should Start Using Today
  
• The algorithm predicts inverted new material compositions
  
• software developers, you're on the wrong track to machine learning
  
• 'White' Artificial Intelligence Risk Increases Racial Inequality, Study Reveals
• How to disable Google Meet in Gmail
• 4 programming languages that will not be completed soon
• Python, R, Programming Languages Thriving Long-Term
• The new AI study may explain why deep learning works