Common diseases
Artificial intelligence found harmful mutations in cows
The staff of Peter the Great St. Petersburg Polytechnic University, using machine learning, revealed fitness-reducing mutations in the DNA of cows.
The used methods of classification of mutations were tested for effectiveness on other domestic animals. The results of the study in the future will help to predict possible genetic problems in small species. The scientific article was published in the journal Evolutionary Applications (Plekhanova et al., Prediction of deleterious mutations in coding regions of mammals with Transfer learning).
Information about the composition of the genomes of various species of living organisms is rapidly accumulating, but man has been and remains the most studied in terms of genetics. It is known which mutations reduce people's resistance to certain diseases or environmental factors. However, there is significantly less information about harmful mutations in the DNA of other mammals and their specific effect on the body.
Russian scientists decided to partially fill this gap by anticipating which changes in the DNA of various animals make them less adapted. To do this, they took several databases on human genomes, in which mutations are divided into different groups (that is, the criteria for classifying mutations were different in each database). Then they "taught" artificial intelligence (AI) to determine the principles by which a particular element of the database belongs to a specific category.
These principles The AI had to transfer information from other data arrays – about the genomes of dogs and mice -to processing. Less is known about harmful mutations in these organisms, but, nevertheless, some data on them are available. Artificial intelligence processed the genomes of mice and dogs in the same way as human ones, "not knowing" which mutation is neutral and which is harmful (however, this was known to researchers).
In most cases, he attributed the changes in DNA to the right category. This meant that the algorithm of such artificial intelligence is also applicable to the analysis of mutations, the significance of which is unknown for the organism. Therefore trained AI was then used to search for malicious DNA changes in the genome of cows. With a high probability, it is the mutations indicated by artificial intelligence as reducing the viability of these animals that are really such.
The new method of searching for mutations that reduce viability will be especially useful for "predicting" them in wild animals whose populations have low numbers. In isolated groups with a small number of individuals, harmful DNA changes accumulate faster, since there is no influx of "good" genes from individuals from the outside. Therefore, small populations of mammals are most vulnerable. Most likely, the degree of their vulnerability will be reduced if mutations that weaken animals are detected in a timely manner. It will be possible to do this by having a special database of genetic data similar to those with which the authors of the study under discussion worked.
Portal "Eternal youth" http://vechnayamolodost.ru