Researchers made bacterial cells that can be taught using algorithms for artificial intelligence

bacterial cells that can be taught using algorithms
Credit: University of Leeds

Researchers made bacterial cells that can be taught using algorithms for artificial intelligence

Technion and MIT scientists worked together to create cells that were created to compute complex functions—sort of “biocomputers”—using concepts from electrical engineering and bioengineering technologies.

Israeli Institute of Technology’s Technion graduate students and researchers The Massachusetts Institute of Technology’s Ron Weiss and Professor Ramez Daniel’s Laboratory for Synthetic Biology & Bioelectronics collaborated to develop genetic “devices” intended to simulate artificial neural circuitry in computation. Recently, Nature Communications published their findings.

A plasmid, a very short DNA molecule that stays distinct from the bacteria’s “natural” genome, was used to introduce the genetic material into the bacterial cell. Plasmids are also found in nature and have a variety of uses. The genetic code of the plasmid was created by the study team to act as a basic computer, or more specifically, a basic artificial neural network. Multiple genes on the plasmid controlled one another’s activation and deactivation in response to external stimuli to achieve this.

The phrase “a cell is a circuit” means what? What makes a computer biological, exactly?

A computer is essentially made up of switches in 0s and 1s. These switches are subjected to operations, such as addition, selection of the maximal or minimal value between them, etc. A computer can play chess or launch a rocket to the moon thanks to more complex operations that rely on the fundamental ones.

The 0/1 switches in the electrical computers we are familiar with are transistors. However, our cells are also computers—just not the same kind. There, a molecule’s presence or absence can function as a switch. Genes generate, change, or remove molecules while activating, triggering, or suppressing other genes. One of the objectives of synthetic biology is to harness these processes, create the switches, and program the genes necessary to enable a bacterial cell to carry out complex functions.

The ability to detect substances and generate organic compounds is built into every cell. The ability to “computerize” certain cellular functions could have significant effects on biomanufacturing and have numerous medical uses.

Under the direction of Prof. Ramez Daniel from the Faculty of Biomedical Engineering at the Technion and in collaboration with Prof. Ron Weiss from the Synthetic Biology Center, MIT, the Ph.D. students (now doctors) Luna Rizik and Loai Danial, along with Dr. Mouna Habib, were motivated by how artificial neural networks operate.

By mixing genetic “pieces,” or altered genes, in innovative ways, they developed synthetic computation circuits and incorporated ideas from neuromorphic electronics into bacterial cells. The end result was the development of bacteria that can be trained using AI algorithms.

The team succeeded in engineering adaptable bacteria that can dynamically switch between reporting the presence of at least one test chemical or two different compounds (that is, the cells were able to switch between performing the OR and the AND functions). Dynamically changing programming allows cells to accomplish a variety of tasks under various circumstances. (Our cells normally do this, in fact.)

The ability to design and manage this process paves the path for more intricate programming, which qualifies the modified cells for more difficult jobs. The scientists were able to create the necessary genetic alterations to the bacterial cells at a substantially lower time and cost thanks to artificial intelligence systems.

Further, the team made use of another characteristic of live cells that is inherent: their capacity to react to gradients. The team was able to use artificial intelligence algorithms to harness this innate ability to create an analog-to-digital converter that can determine whether a molecule’s concentration is “low,” “mid,” or “high.” Such a sensor could be utilized to provide medications at the right dosage, including cancer immunotherapy and diabetes medications.


Luna Rizik et al. (2022). Synthetic neuromorphic computing in living cells, Nature CommunicationsDOI: 10.1038/s41467-022-33288-8

Related Links:

Research Summary: One Bacterial Cell, One Complete Genome

ABSTRACT While the bulk of the finished microbial genomes sequenced to date are derived from cultured bacterial and archaeal representatives, the vast majority of microorganisms elude current culturing attempts, severely limiting the ability to recover complete or even partial genomes from these environmental species. Single cell genomics is a novel culture-independent approach, which enables access … Continue reading

Research Summary: Bacterial Programmed Cell Death and Multicellular Behavior in Bacteria

Abstract Traditionally, programmed cell death (PCD) is associated with eukaryotic multicellular organisms. However, recently, PCD systems have also been observed in bacteria. Here we review recent research on two kinds of genetic programs that promote bacterial cell death. The first is mediated by mazEF, a toxin–antitoxin module found in the chromosomes of many kinds of … Continue reading