A new neural network predicts the growth and formation of DNA

Presents a computer program that uses artificial intelligence for DNA analysis and prediction of complex human traits, including growth, susceptibility to osteoporosis and the expected level of education.

Новая нейросеть предсказывает рост и образование по ДНК

Main scientific achievements 2016

American scientists have proposed an algorithm that is able to interpret DNA profiles and to predict growth and bone density, and even the education level of the person. “Although we confirmed the efficiency of the tool for these three specific cases, we can now apply the same method for the prediction of other complex characteristics associated with risk of cardiovascular disease, diabetes and breast cancer,” says Stephen Hsu (Stephen Hsu) of the University of Michigan, lead author of an article published in the journal Genetics.

In their work the researchers used anonymized personal data and DNA sequencing to 500 thousand inhabitants of great Britain, contained in an open database of UK Biobank. Was prepared by a neural network capable of analyzing single-nucleotide substitutions, there may be only minimal differences in the genomes of different people. After the necessary training she could really predict the height of a man, according to DNA up to a couple inches.

The accuracy of the prediction of education level and bone density was not so high, however, the options — the risk of developing abnormally fragile bones or serious difficulties even with the school teaching the neural network predicted confidently. However, the authors continue to Refine their program and are working on expanding the dataset to train the neural network. This should improve the results and expand the application of the new tool, which someday may become an important part of future medicine.