Appr header

A unified
platform
driven by cell
behavior

Innovation in drug discovery doesn’t need a renovation. It needs a rearchitecture.

Cellarity has developed a broad platform that generates proprietary, high content, single-cell data and uses cutting-edge machine learning algorithms to interpret the n-dimensional complexity of biology. These capabilities allow Cellarity to perceive the network-state of a given cell—defining a cell’s behavior—and enable the platform to unlock the ability to study cellular behaviors, unravel the network dynamics that govern those behaviors and generate medicines that can direct them. This endows the platform with the unique ability to design drugs against cellular behavior.

The Cellarium™

Cellarity’s approach is to embrace clinical hypotheses generated from AI predictions, available data or biological insights.

Clinical programs are iterated rapidly via an interdisciplinary interpretation, curation and learning model.

Because of this speed, Cellarity is able to generate medicines against a wide breadth of therapeutic areas at an unprecedented throughput.

Cellarium Platform Diagram desktop 20191202 4x 8 Cellarium Platform Diagram mobile 20191202 4x 8

Cellarity’s drug discovery platform is built on two integrated components: Our physical laboratory, equipped with the latest in biology, chemistry, gene technologies, sequencing, robotics and 3D printing, and its digital twin, the Cellarium, a computational platform that uses artificial intelligence to translate fit-for-purpose wet lab data into digital biology.

Cellarity Maps™—cellular and drug modality maps—allow for cross-species, cross-disease and cross-therapeutic interpretation of cell behavior and, importantly, provide a therapeutic knowledge graph that supports rapid designing, testing and validation of new clinical hypotheses.

The Cellarium uses an elegant and efficient network of symbolic and non-symbolic machine learning approaches to chart new clinical hypotheses across Cellarity Maps — digital guides that enable our biologists and machine learning scientists to predict, explore and navigate cell behaviors. A library of Cellarity Maps is cataloged in the Cellarity Atlas™, providing a global view of biology. Cellarium-generated therapeutic predictions are then validated in wet lab analyses before moving to the clinic.

An Interdisciplinary, Non-Hierarchical Approach

Group 22 Group 23 Group 21

The Cellarity Technology team generates the data that drive the Cellarium™ prediction engine and creates and implements advanced data modalities that will yield richer predictions. The technology team:

  • Builds, operates and optimizes rigorously managed, cutting-edge processes to generate high quality and fit-for-purpose data, including both industry standard and proprietary processes.
  • Explores and innovates novel and powerful methods and technologies to create better or complementary datasets that enrich the Cellarium.

The Cellarity AI team explores therapeutic approaches of the discovery arm of the Cellarium. The AI team:

  • Identifies, infers and generates clinical hypotheses for cellular behaviors by applying sophisticated, sensible, benchmarked and reproducible machine learning methods on fit-for-purpose data.
  • Generates new molecules and predicts pharmacological perturbations that both maximize on-target effects and minimize off-target effects, thus accelerating translation into clinical practice.

The Cellarity Biology team powers and leverages the Cellarium prediction engine, generating and testing hypotheses in the physical lab to identify new medicines. The biology team:

  • Develops cutting-edge laboratory models to generate the biological and chemical data that drive digital predictions.
  • Validates digital predictions and identifies drug leads for preclinical development.
  • Group 22

    Technology

    The Cellarity Technology team generates the data that drive the Cellarium™ prediction engine and creates and implements advanced data modalities that will yield richer predictions. The technology team:

    • Builds, operates and optimizes rigorously managed, cutting-edge processes to generate high quality and fit-for-purpose data, including both industry standard and proprietary processes.
    • Explores and innovates novel and powerful methods and technologies to create better or complementary datasets that enrich the Cellarium.
  • Group 23

    Artificial Intelligence

    The Cellarity AI team explores therapeutic approaches of the discovery arm of the Cellarium. The AI team:

    • Identifies, infers and generates clinical hypotheses for cellular behaviors by applying sophisticated, sensible, benchmarked and reproducible machine learning methods on fit-for-purpose data.
    • Generates new molecules and predicts pharmacological perturbations that both maximize on-target effects and minimize off-target effects, thus accelerating translation into clinical practice.
  • Group 21

    Biology

    The Cellarity Biology team powers and leverages the Cellarium prediction engine, generating and testing hypotheses in the physical lab to identify new medicines. The biology team:

    • Develops cutting-edge laboratory models to generate the biological and chemical data that drive digital predictions.
    • Validates digital predictions and identifies drug leads for preclinical development.

Drug discovery largely follows the same approach used at the turn of the century. The focus is on isolated genes and pathways rather than on disease-causing cell behaviors.

Work proceeds linearly from a specific scientific hypothesis through data generation and then to analysis, with each step carried out by separate groups. When machine learning is employed, it is often only as an afterthought at the end of a given translational process or as a demonstration project run in parallel, and genomics data generation is handled with an outsourcing model. The end result is a lack of synergy and integration, leading to widespread missed opportunities.

The technology and capability to transform drug discovery exist now.

Cellarity is seizing the moment by leveraging existing and developing new state-of-the-art genomics and machine learning technologies to power a new, cell behavior-centric paradigm for drug discovery.

The integrated, non-hierarchical structure of Cellarity provides a critical advantage. Biology, Technology and AI operate through a nonlinear approach with a single unified vision. By synergistically blending these capabilities, Cellarity maximizes the efficiency and effectiveness of its approach. It breaks through barriers faced by traditional drug discovery companies and overcomes challenges that new AI companies face in both fit-for-purpose data and in siloed application of AI to traditional processes.

Rearchitecting Drug Research and Development

1

Unifying
Biology


As the fundamental biological unit of all organisms, cells contribute to human health through their proper function and organization. The behavior of cells is determined by complex molecular networks, where diseases can emerge when these networks become dysregulated. Therefore, medicines targeting molecular networks will be more efficacious. Learning the full molecular lexicon and grammar of cellular behaviors will enable treatment of complex diseases.

2

Directing
Cell Behaviors


While a single word can be powerful, phrases or sentences often have greater impact. Similarly, while some diseases arise from a single dysfunctional protein, many diseases arise from complex molecular networks. While the networks underlying cell behaviors cannot be easily deciphered by humans, machines are adept at retaining and harnessing complexity. Medicines targeting these networks will be highly potent and selective.

3

Blending
Disciplines


Only companies that speak the languages of biology, technology and computation equally fluently and without hierarchical relationships will be able to frame biological problems in machine-understandable ways and improve the way machines learn from data, enabling AI to generate the most predictive insights.

4

Reinforcing
Understanding


The path from disease to data to insight is non-linear and iterative, with multiple points that require human interpretation and testing. The combination of scientists with machines will outperform scientists or machines alone.

5

Embracing
Complexity


Experiments must be designed to maximize a computer’s ability to extract the molecular drivers of disease. More than merely collecting large datasets and applying AI as an afterthought, embracing complexity is an essential requirement to maximize the power of data. This approach forgoes standard lab-adapted models in favor of systems from which the most relevant and useful governing principles can be learned. Hypotheses are knowledge- or data-driven, further refined or evolved by fit-for-purpose data.