Machine learning can be predicted from genomic repair results

Meng ran Zhang

In November 2018 09, 08:21 source: Science Daily
Original title: machine learning can predict genome repair results

The UK online magazine "nature" published on the 8 an artificial intelligence and biotechnology research, scientists reported a method of pathogenic gene mutation to achieve precise and predictable editing by machine learning. The results for the study of genetic diseases, develop potential therapy provides a new possibility.

Although CRISPR-Cas9 has revolutionized the study for genome editing technology, but at present, in order to ensure the accuracy of this technique is especially important.

CRISPR-Cas9 genome editing commonly used DNA "template" to ensure the accuracy of DNA repair, or the specific DNA sequences into the genome. Therefore, the lack of DNA repair of these templates is often considered accurate enough.

Now, the United States Brigham and women's Hospital and Harvard Medical School scientists Richard Schulwood and colleagues developed a learning method to predict the results of the genome repair machine, to achieve the precise template free Cas9 editor. The research team used a contains nearly 2000 of Cas9 RNA (gRNA) and human wizard DNA target database, training a inDelphi machine learning model.

Through the model identification, 5% Cas9 RNA to 11% of the human genome wizard can target in more than 50% of the cases (known as "precision -50") produce repair results of single and predictable. InDelphi can use the template free Cas9 editor to identify and predict pathogenic gene mutation appropriate target, including some considered cannot be found by the method of target.

Finally, the research team proved by experiments, the cells of the body and Pudelake hermanski - syndrome (HPS, a kind of albinism syndrome), Menkes disease (also known as gray hair dystrophy) and familial hypercholesterolemia of these three diseases nearly 200 species of Pathogenic Differentiation, the editing and repair accuracy to achieve accurate -50 standard.

The researchers say the results establish a precise method without template, genome editing. (Zhang Mengran)

(commissioning editor Liu Jingting and Xiong Xu)

Recommended reading

Guide to living away from cancer Recently, many public figures had died from cancer. The memorial, I sigh, prevention and treatment of cancer topics once again become the focus of public discussion. The people suffering from cancer is more and more? What are the high-risk population of cancer? Cancer prevention need to pay attention to what the problem? The risk of cancer big people have what. [details]

The Milky Way genetic data secret Gaia telescope According to a study published in the British Astronomical Journal "nature" on the 1 day, the European team of scientists released by Gaia telescope data, determine where we formed the Milky Way -- about 10 billion years ago, the Milky Way system with another galaxy collision, resulting in the Milky Way Department of the halo is full of stars, and the silver thickening. The colliding galaxies thus found are named. [details]