Biologists are currently trying to use artificial intelligence to create virtual human cells. This is an alternative to complex computer models.
In particular, one of the leaders of the Chang Zuckerberg Initiative Research Project in California. Redwood City Stephen Quake points out that the goal is to create a computational tool for transition cell biology from 90% of experiments to 90% of calculations. Currently, artificial intelligence models can process and generate information that can take weeks to experimentally validate, such as modeling the response of cancer cells to a particular drug.
«It can be a very powerful tool to understand what is bad in treating disease. Scientists can experiment with virtual cells to see predictions» explains Stephen Quake.
Creating virtual cells has already sparked widespread interest among researchers. The Chan Zuckerberg initiative is planning to invest hundreds of millions of dollars in creating virtual cells over the next decade.
Meanwhile, Google Deepmind CEO Demis Hasabis confirmed earlier this year that his company is also working on a project to develop virtual cells. There is a similar project in Sweden. There, they are working on a unique virtual cell model called Alpha Cell, which is scheduled to be released in 2026.
However, many scientists have pointed out that the rush of creating virtual cells causes a lot of hype, but in reality there are no clear successes or concrete results. Biologists have been using computer models for decades to predict cell behavior. In 2012, scientists created the first computational model of the whole cell, reflecting the inner behavior of the bacterial mycoplasma genitals, which only has 525 genes.
The current approach to virtual cell development is to use advances in artificial intelligence to generate more complex cellular mechanisms. Early attempts to create virtual cells were based on the use of only one type of data that was experimentally obtained as part of Sequencing of all RNA molecules in individual cells. These data formed a momentary record of the basis of the gene activity catalogue and current cellular state.
Creators – Global PR Agency and B2B Companies for Technology
PR services for businesses and their executives Media interviews, podcast appearances, and meeting presentation arrangements Europe, Asia, AmericaCatalogs are used to create atlases that display different types of cells in the human body. Currently, researchers are using these datasets to create virtual cells using AI. The Chan Zuckerberg initiative is set to release data on roughly 1 billion cells. In particular, the Arc Institute recently launched its first virtual cell model called State.
Meanwhile, some researchers point out that such models are not strong enough to make clear predictions that can be used beyond the trained data. According to most researchers, virtual cell models should also include other types of data, such as images taken with optical and electron microscopes. Such data allow AI-based models to understand how different components within a cell interact with each other, and how the cells themselves change over time.
Furthermore, there is no clear definition of virtual cells yet, as concepts can include different things depending on the needs of different people. However, scientists hope that truly effective AI models that can control virtual cells could emerge in the near future.
Source: Nature
