We build virtual cell models for Turbine.

We are the research arm of Turbine, focused on understanding and predicting cell behavior beyond what current experiments and models allow. This site is where we share how we think about the state of the field, publish what we learn and invite others to work with us on key problems.

Our thinking has been shaped by building computational models of perturbation biology (Virtual Cells) for drug discovery within Turbine for the last decade. Our models are successful if and only if they help us understand biology better and in turn manage to influence actual decisions in drug discovery. This gives us a different worldview from purely academic groups.

We are working towards a general model of biology.

Identical twins show that natural biology reproduces very accurately. Lab experiments may struggle measuring or reproducing certain effects, but we believe biological programs defined in the DNA of organisms exist and will eventually be learnable to a degree where we can accurately predict effects beyond what any lab lets us observe. These would be the general Virtual Cells - almost like an "AGI of biology".

The state-of-the art today is Virtual Assays - narrow virtual cells of existing assays.

We see that the predictivity of the cell models of today are bounded by the circumstances in which the backing data was generated. Current models can generalize to new biological models, even some new perturbations within an assay, but every single assay to be predicted needs to be grounded individually in data. This is why we call the models released at Turbine Virtual Assays.

We advance the field by connecting diverse data instead of generating big homogeneous data.

To bridge that gap from narrow to general Virtual Cells, we think the missing piece is to understand how the cells behave in different circumstances as opposed to generating one large homogeneous data set with a single protocol.

Therefore we specialize in understanding how to make different types of data work together and integrating these seemingly disparate data types into a single, coherent architecture. Turbine's wet-lab also specializes in agile data generation, being able to spin up new ML-grade seed data sets in a month rather than focusing on one single data type.

Pushing the limits of understanding biology is a shared effort.

We think furthering computational biology is best done as a combined effort, not in individual silos. We aim to share our findings and thinking quickly, with all the rough edges, but as most interesting results rarely add up to a nice story, we frequently use non-traditional publication formats.

We are happy to collaborate if you share our pains (see Open problems) or have an interesting dataset which could use our integration chops.