Precigenetics unveils Cell Cinema, which turns living cells into time series for AI

Precigenetics has shown how the same living cell can be measured repeatedly after drug treatment

On July 21, Precigenetics presented a technical demonstration of Cell Cinema. The system repeatedly measures the same living cell after drug treatment and converts the sequence of observations into a data series for models.

Many cell experiments provide only an endpoint. The cells are fixed or destroyed, and the state of the population is measured after treatment. Such a snapshot can reveal differences among cells, but it cannot show the sequence of changes within a particular cell.

Cell Cinema keeps the cell alive and repeatedly captures three-dimensional images without staining labels. Each pixel in an image contains a broad spectrum of light. The software converts each measurement into a set of numbers that describes the cell's state. A sequence of these sets provides a record of changes in the same cell before, during, and after treatment.

In its first public demonstration, the company showed SK-MEL-2 melanoma cells treated with RSL3. This drug inhibits the GPX4 protein and triggers ferroptosis, a form of cell death caused by iron-dependent lipid oxidation. In a second experiment, vemurafenib altered the chemical state of A375 cells carrying a BRAF mutation. The page includes processed images and the resulting records of how the cells changed over time.

A record of one cell replaces a snapshot of a cell population. After drug treatment, the system records a sequence of measured states for that cell. The company expects these records to help identify early signs of toxicity, compare doses, and select the next experiment based on the cell's observed response.

In the spring, MaxToki reconstructed an age-related trajectory from single measurements of approximately 175 million cells. Cell Cinema provides a different type of material for similar tasks: repeated observations of the same cell instead of comparisons among different cells sampled at different points.

The company proposes validating these records through prediction: the early part of a cell's history should predict its later state in cells that were not used for training. The company also proposes two further tests: preserving the description of a cell's state across runs and transferring it to new conditions. In this way, Cell Cinema compares its measurement series with a single snapshot and with analysis performed after the experiment.

Age-related damage and responses to treatment also develop over time. Precigenetics plans to use these records so that models can select the next experiment from a living cell's early response. For intervention studies, this leads to a practical question: which early change in the cell's history should be tested next with another drug?