Lightning.AI Platform Technologies

Thinking Pathways

Pathways Hypotheses

Leveraging knowledge built over a decade, the Lightning.AI integrates literature and omics with multiple proprietary sources such as curated publications, analyses from proprietary dataset of over 2 billion visual biology images and thousands of experimental results from our projects, across diseases, targets, and cell types.

Thousands of pathway hypotheses are generated and then seen by Experimental AI in the next iteration.


2 Billion Visual Biology images
Lightning.AI dataset
Omics
Literature
Thousands of
experimental results
Curated publications

Thinking Pathways

Literature and Omics

Literature and multi-omics spark ideas and give the initial input to generate thousands of hypotheses, but our disease-specific proprietary data from more than 2B images and self-learning neural networks is what turns them into breakthroughs.


Speaking Pathways

Lightning.AI Co-Pilot

The Lightning AI Co-Pilot provides a ChatGPT-like interface to “chat with cells”, enabling biologists, chemists, and data analysts to interact with the system in a seamless user experience.

Using the LLM, the Co-Pilot translates questions in natural language into queries that are saved along with their results, with each step further elucidating disease biology. It also acts as a way to ask the system to plan, execute, and monitor discovery activities.


Speaking Pathways

Pathway LLM

Built to connect AI with experimental biology, LLM enables “chat with cells”. It empowers AI reasoning with real disease biology, enabling us to see, validate and learn biology from real cells, through functional, experimental reasoning at AI scale.


Seeing Pathways

PathwayLight

PathwayLight technologies visualize how thousands of pathways behave inside real cells. These technologies including TranslationLight, TranscriptLight and others, see regulatory events and pathways such as splicing, localization, modification, or decay/stability, creating visualizers. Each visualizer is built as a high-content screening assay with the images acting as visual readouts of pathway biology.

  • Transcription
  • Splicing
  • Alternative splicing
  • Capping
  • Polyadenylation
  • Modification
  • Nuclear Export
  • Transport
  • Localization
  • Stability
  • Decay
  • Translation

Seeing Pathways

Automated Visual Biology Lab

Our lab runs high-content at high-throughput screening in a massively parallel architecture. With a screen-to-cloud architecture, billions of images are uploaded to our cloud servers for analysis. The fully automated lab uses robotics and high-resolution microscopy, incorporating our imaging and assays under a single framework.


Learning Pathways

PathwayLight Neural Network

Trained on billions of images, this AI inside the biology compares healthy and diseased cells millions of times for every hypothesis. It identifies dysregulation, recognizes phenotypes, and drives the discovery loop forward. It feeds that insight back to discovery loop - iteratively converging on novel, druggable targets. The final outcome is the discovery of the targets directly linked to disease biology with biological proof at each step.


Diseased
Phenotype
Healthy
Phenotype

Learning Pathways

mRNA Biology Neural Network

Trained on mRNA biology regulation across thousands of pathways, the second neural network explores each target’s mRNA biology. It identifies modulation strategy through splicing, RBPs, localization, degradation or translation. As compounds are screened, the neural network visualizes their effect on disease pathways in real cells - confirming impact and optimizing intervention. The final outcome is the discovery of the drugs using mRNA modulation with mechanisms of actions as a strategy to drug even the undruggable targets.