Generative AI Fills In The Gaps In Microscopy Data To Further Genetic Medicine
Press Release – Skoltech
DNA requires more than the right set of genes: It has to have the correct 3D architecture, which is traditionally the object of statistical physics, and polymer physics in particular.
Skoltech researchers have enlisted generative artificial intelligence to complete the missing data on the distances between pairs of genes in DNA. This enables figuring out the 3D architecture of DNA molecules, which is in turn necessary for developing treatments and diagnostic approaches for genetic diseases. Published in the journal Scientific Reports, the study is the first successful attempt to flesh out such data using AI or, in fact, by any means. Previously, scientists had to make do with incomplete data, hampering progress in medical genetics and limiting the scientists' understanding of the biophysics of chromatin — the stuff of chromosomes.
To do its job properly, DNA requires more than the right set of genes: It has to have the correct 3D architecture, which is traditionally the object of statistical physics, and polymer physics in particular. The way the 46 long DNA macromolecules per cell are folded in space affects which genes are active and whether the cell will reproduce appropriately and differentiate into specialized cell types during embryonic development. Conversely, faulty DNA architecture plays a role in the development of abnormalities and diseases, such as cancer.
The more scientists learn about the physical principles behind the stabilization of the 'healthy' 3D architecture of DNA, the more opportunities for diagnosing and treating genetic disorders are created. By comparing DNA spatial structure in health and disease, biomarkers for diagnosing disorders and personalized treatments can be found. Scientists can identify new therapeutic targets, develop drugs that restore normal gene function, and design precise gene editing interventions.
One of the most widely used experimental techniques for examining how DNA molecules are folded in space is fluorescence microscopy. This refers to a kind of optical microscopy where certain specific gene sequences — a great number of those, in fact — are highlighted by staining them with fluorescent tags.
The problem is that such data is inevitably fragmentary. To attach a fluorescent tag, scientists synthesize a short gene sequence that is complementary to the sequence at the position of interest along the DNA strand. However, it's not possible for every sequence. If it contains repeated nucleobases, such as a string of letters A, for example, the sequence cannot be stained selectively, because it is not unique. So researchers have had to make do with incomplete data. Not anymore.
'Once you know the distances between a sufficient number of genes, determining the remaining distances for which there is no experimental data takes the form of a mathematical problem with a specific solution,' the principal investigator of the study, Assistant Professor Kirill Polovnikov from Skoltech Neuro, commented. 'We have shown for the first time that generative models are capable of solving such problems. This is an unconventional application of the kind of AI usually employed for more 'creative' tasks — generating images and text based on a user prompt. At the same time, this is a new approach to the study of chromatin structure, where polymer physics has historically reigned supreme.'
The implications of the research are twofold. Practically speaking, the Skoltech team has proposed and tested a way to process fluorescent microscopy data that will ultimately enable a better understanding of DNA spatial structure, which promises better treatments and diagnostics for genetic diseases. Fundamentally, the study demonstrates the potential of generative artificial intelligence beyond the usual scope of its applications.
Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
Scoop
6 hours ago
- Scoop
Generative AI Fills In The Gaps In Microscopy Data To Further Genetic Medicine
Press Release – Skoltech DNA requires more than the right set of genes: It has to have the correct 3D architecture, which is traditionally the object of statistical physics, and polymer physics in particular. Skoltech researchers have enlisted generative artificial intelligence to complete the missing data on the distances between pairs of genes in DNA. This enables figuring out the 3D architecture of DNA molecules, which is in turn necessary for developing treatments and diagnostic approaches for genetic diseases. Published in the journal Scientific Reports, the study is the first successful attempt to flesh out such data using AI or, in fact, by any means. Previously, scientists had to make do with incomplete data, hampering progress in medical genetics and limiting the scientists' understanding of the biophysics of chromatin — the stuff of chromosomes. To do its job properly, DNA requires more than the right set of genes: It has to have the correct 3D architecture, which is traditionally the object of statistical physics, and polymer physics in particular. The way the 46 long DNA macromolecules per cell are folded in space affects which genes are active and whether the cell will reproduce appropriately and differentiate into specialized cell types during embryonic development. Conversely, faulty DNA architecture plays a role in the development of abnormalities and diseases, such as cancer. The more scientists learn about the physical principles behind the stabilization of the 'healthy' 3D architecture of DNA, the more opportunities for diagnosing and treating genetic disorders are created. By comparing DNA spatial structure in health and disease, biomarkers for diagnosing disorders and personalized treatments can be found. Scientists can identify new therapeutic targets, develop drugs that restore normal gene function, and design precise gene editing interventions. One of the most widely used experimental techniques for examining how DNA molecules are folded in space is fluorescence microscopy. This refers to a kind of optical microscopy where certain specific gene sequences — a great number of those, in fact — are highlighted by staining them with fluorescent tags. The problem is that such data is inevitably fragmentary. To attach a fluorescent tag, scientists synthesize a short gene sequence that is complementary to the sequence at the position of interest along the DNA strand. However, it's not possible for every sequence. If it contains repeated nucleobases, such as a string of letters A, for example, the sequence cannot be stained selectively, because it is not unique. So researchers have had to make do with incomplete data. Not anymore. 'Once you know the distances between a sufficient number of genes, determining the remaining distances for which there is no experimental data takes the form of a mathematical problem with a specific solution,' the principal investigator of the study, Assistant Professor Kirill Polovnikov from Skoltech Neuro, commented. 'We have shown for the first time that generative models are capable of solving such problems. This is an unconventional application of the kind of AI usually employed for more 'creative' tasks — generating images and text based on a user prompt. At the same time, this is a new approach to the study of chromatin structure, where polymer physics has historically reigned supreme.' The implications of the research are twofold. Practically speaking, the Skoltech team has proposed and tested a way to process fluorescent microscopy data that will ultimately enable a better understanding of DNA spatial structure, which promises better treatments and diagnostics for genetic diseases. Fundamentally, the study demonstrates the potential of generative artificial intelligence beyond the usual scope of its applications.
Scoop
8 hours ago
- Scoop
Generative AI Fills In The Gaps In Microscopy Data To Further Genetic Medicine
Skoltech researchers have enlisted generative artificial intelligence to complete the missing data on the distances between pairs of genes in DNA. This enables figuring out the 3D architecture of DNA molecules, which is in turn necessary for developing treatments and diagnostic approaches for genetic diseases. Published in the journal Scientific Reports, the study is the first successful attempt to flesh out such data using AI or, in fact, by any means. Previously, scientists had to make do with incomplete data, hampering progress in medical genetics and limiting the scientists' understanding of the biophysics of chromatin — the stuff of chromosomes. To do its job properly, DNA requires more than the right set of genes: It has to have the correct 3D architecture, which is traditionally the object of statistical physics, and polymer physics in particular. The way the 46 long DNA macromolecules per cell are folded in space affects which genes are active and whether the cell will reproduce appropriately and differentiate into specialized cell types during embryonic development. Conversely, faulty DNA architecture plays a role in the development of abnormalities and diseases, such as cancer. The more scientists learn about the physical principles behind the stabilization of the 'healthy' 3D architecture of DNA, the more opportunities for diagnosing and treating genetic disorders are created. By comparing DNA spatial structure in health and disease, biomarkers for diagnosing disorders and personalized treatments can be found. Scientists can identify new therapeutic targets, develop drugs that restore normal gene function, and design precise gene editing interventions. One of the most widely used experimental techniques for examining how DNA molecules are folded in space is fluorescence microscopy. This refers to a kind of optical microscopy where certain specific gene sequences — a great number of those, in fact — are highlighted by staining them with fluorescent tags. The problem is that such data is inevitably fragmentary. To attach a fluorescent tag, scientists synthesize a short gene sequence that is complementary to the sequence at the position of interest along the DNA strand. However, it's not possible for every sequence. If it contains repeated nucleobases, such as a string of letters A, for example, the sequence cannot be stained selectively, because it is not unique. So researchers have had to make do with incomplete data. Not anymore. 'Once you know the distances between a sufficient number of genes, determining the remaining distances for which there is no experimental data takes the form of a mathematical problem with a specific solution,' the principal investigator of the study, Assistant Professor Kirill Polovnikov from Skoltech Neuro, commented. 'We have shown for the first time that generative models are capable of solving such problems. This is an unconventional application of the kind of AI usually employed for more 'creative' tasks — generating images and text based on a user prompt. At the same time, this is a new approach to the study of chromatin structure, where polymer physics has historically reigned supreme.' The implications of the research are twofold. Practically speaking, the Skoltech team has proposed and tested a way to process fluorescent microscopy data that will ultimately enable a better understanding of DNA spatial structure, which promises better treatments and diagnostics for genetic diseases. Fundamentally, the study demonstrates the potential of generative artificial intelligence beyond the usual scope of its applications. 25-13-00277.
Scoop
9 hours ago
- Scoop
Neural Networks Speed Up Search For Solid-State Battery Materials For Safer Electric Cars With Extended Range
Press Release – Skoltech So far solid-state batteries have not been adopted by carmakers, but EV developers are looking to capitalize on the technology before competitors. The new type of energy storage could improve fire safety and boost EV range by up to 50%. Researchers from Skoltech and AIRI Institute have shown how machine learning can speed up the development of new materials for solid-state lithium-ion batteries. These are an emerging energy storage technology, which could theoretically replace conventional Li-ion batteries in electric vehicles and portable electronics, reducing fire hazards and extending battery life. In the Russian Science Foundation-backed study, published in npj Computational Materials, neural networks proved capable of identifying promising materials for the key component of these advanced batteries — the solid electrolyte — as well as for its protective coatings. Like its conventional counterpart, the solid-state battery incorporates an electrolyte, through which ions carrying the electric charge travel from one electrode to another. While in a conventional battery the electrolyte is a liquid solution, its solid-state analogue, as the name suggests, relies on solid electrolytes, such as ceramics, to conduct lithium ions. So far solid-state batteries have not been adopted by carmakers, but EV developers are looking to capitalize on the technology before competitors. The new type of energy storage could improve fire safety and boost EV range by up to 50%. The problem is that none of the currently available solid electrolytes meets all the technical requirements. So the search for new materials continues. 'We demonstrated that graph neural networks can identify new solid-state battery materials with high ionic mobility and do it orders of magnitude faster than traditional quantum chemistry methods. This could speed up the development of new battery materials, as we showed by predicting a number of protective coatings for solid-state battery electrolytes,' commented the lead author of the study, Artem Dembitskiy, a PhD student of Skoltech's Materials Science and Engineering program, a research intern at Skoltech Energy, and a junior research scientist at AIRI Institute. Study co-author, Assistant Professor Dmitry Aksyonov from Skoltech Energy explained the role of protective coatings: 'The metallic lithium of the anode is a strong reducing agent, so almost all existing electrolytes undergo reduction in contact with it. The cathode material is a strong oxidizing agent. When oxidized or reduced, electrolytes lose their structural integrity, which can degrade performance or even cause a short circuit. You can avoid this by introducing two protective coatings that are stable in contact with the anode and the electrolyte and the cathode and the electrolyte.' Machine learning algorithms make it possible to accelerate the calculation of ionic conductivity, a key property both for electrolytes and for protective coatings. It is among the most computationally challenging characteristics calculated in screening the candidate materials. For protective coatings, the list of properties that are checked at various stages of material selection includes thermodynamic stability, electronic conductivity, electrochemical stability, compatibility with electrode and electrolyte materials, ionic conductivity, and so on. Such screening happens in stages and gradually narrows down the list of perhaps tens of thousands of initial options to just a few materials. The authors of the study used their machine learning-accelerated approach to search for coating materials to protect one of the most promising solid-state battery electrolytes: Li10GeP2S12. The search identified multiple promising coating materials, among them the compounds Li3AlF6 and Li2ZnCl4.
