About X-Tour
We are a biotechnology company interested in AI tools and simulation pipelines for protein engineering. Our interests sit at the intersection of machine learning and molecular biophysics, where predictive models meet physical simulation.
Who We Are
Our interests span the full computational stack for protein science: from models that predict protein properties out of sequence and structure, to simulations that show the molecular behaviour behind those predictions.
We believe the most impactful tools in computational biology come from combining data-driven approaches with physics-based methods. Machine learning captures patterns in large datasets; molecular simulations provide the mechanistic insight to understand why those patterns exist. We are interested in tools that bring both together.
We are interested in computational tools that are useful in protein engineering.
What We Believe
Models should know their limits
In protein science, a confident but wrong prediction can send research down the wrong path for months. Uncertainty quantification, knowing where a model holds up and where it doesn't, is part of what interests us in any model.
Physics and data are complementary, not competing
Pure ML models can learn spurious correlations. Pure physics is computationally expensive. The best tools in computational biology use both: data-driven pattern recognition grounded in physical principles.
Models point to the experiments worth running
A predictive model does not replace an experiment, it generates hypotheses: it narrows the space of possible variants down to the few worth testing in the lab. The smaller that space, the faster new knowledge appears.