How is machine learning used in genomics

What AI can (and cannot) do in genome research

Technology in the genes

The decoding of the human genome is still a mystery. Artificial intelligence could help solve this. New therapeutic approaches for serious diseases seem possible, as well as non-medical “improvements” in the genetic make-up.

Technology assessors from the Karlsruhe Institute of Technology (KIT), funded by the Federal Ministry of Education and Research (BMBF), examine which applications are realistic and which ethical questions they pose to society. "Modern genome research wants to understand and predict how genetic differences between people determine complex characteristics, such as dispositions for common diseases," says Harald König from the Institute for Technology Assessment and Systems Analysis (ITAS) at KIT. Although the possibilities of analyzing the genetic material are advancing rapidly, the knowledge about how our genetic material determines such characteristics has so far mostly been limited to correlations. The use of advanced forms of machine learning now promises a decisive further development.

Effective therapies against cancer or dementia

The result could be an enormous leap in knowledge - from correlations to causal relationships - which promises completely new application possibilities

Harald King

The new approaches rely on the combination of artificial intelligence and rapidly advancing techniques of genome analysis (including single cell analysis) as well as automated laboratory platforms. The latter can provide very large amounts of data on genome changes and various cellular processes such as the reading of genes or the appearance of different protein forms under different conditions. "The result could be an enormous leap in knowledge - from correlations to causal relationships - that promises completely new application possibilities," says Harald König. Such approaches are linked, for example, to the hope of new, far more effective therapeutic approaches for cancer, cardiovascular diseases or dementia. This “precision medicine” could be used tailor-made for different groups of patients, disease variants or stages.

The research team, in which the Fraunhofer Institute for Systems and Innovation Research ISI is involved in addition to the technology assessment of KIT, wants to analyze which applications are realistic in practice in the short to medium term. At the same time, her focus is on the diverse social and political implications with which the new knowledge is connected. The medicine of the future could have enormous macroeconomic and social benefits for an aging society. At the same time, however, some of their approaches, such as gene- and cell-based therapies, could also be associated with very high costs, which raise questions about funding for research and development and accessibility for patients.

Preventive intervention in the human germline

The knowledge of which genetic information would have to be "rewritten" and how in order to achieve certain effects, together with the most recent methods of genome editing such as the CRISPR-Cas system, also raises ethical questions. Thus, at least internationally, the trend could go beyond rare hereditary diseases and also prevent common ailments such as breast cancer or diabetes by “preventive correction” of corresponding risk mutations in the germline of human embryos. "A development which in extreme cases could lead to growing acceptance of 'improving' the human genome with non-medical interventions," explains Harald König.

In addition, society has to deal with how the property rights of genetic data can be regulated and their security guaranteed. In the future, for example, it may be possible to use genome sequences to directly draw conclusions about the phenotype, e.g. the appearance of people. "This knowledge," said König, "would not only be extremely valuable for law enforcement authorities."

You might also be interested in this article

Time for a revolution?

From the end of medicine as we know it ...

Many researchers and experts are currently assuming that the next great socio-economic revolution will involve a complete redefinition of health and how we define and treat diseases. “Our current health system can no longer be maintained as it is. It has become too expensive and too ineffective, "says Professor Harald Schmidt, ...

Policy Makers Options

Based on their technology assessment, the researchers want to develop options for research and innovation policy over the next two years. They also want to offer political decision-makers various options for action that open up ways for a responsible use of this technology, which is driven by AI and genomics.

Source: Karlsruhe Institute of Technology (KIT)