Platform
Platform
The immune system is at the center of human health and disease.
Overview
Immunai leverages big data, single cell multi-omics, and ML to bridge the gap between causal immunology and translational-disease biology. We generate cell-specific perturbation signatures across multiple cohorts in order to inform drug discovery and development decisions.
Identify discovery cohorts
Inflammatory diseases and cancer
Reverse translational discovery
Identify systems-level immune dysregulation
Target validation
Nominate and validate targets to reverse imbalance
Therapeutics development
Development of agonist and antagonist modalities
Treatment cohorts
Inflammatory diseases and cancer
Solutions
Immunai’s platform enables novel, data-driven insights that reveal actionable therapeutic targets and accelerates drug development for immune-modulating medicines.
Immunai links causal experimental biology and observational translational insights
Data generation and curation
ML-driven target ID and prioritization
Validation
AMICA™ – the world’s largest cell-level immune knowledge base
Immunai’s rich clinico-genomic data assets are harmonized in AMICA™, the – Annotated Multi-Omic Immune Cell Atlas. Fed by massive scale curated public -omics data and proprietary cohorts and experiments, AMICA™ covers 10s of millions of cells across hundreds of disease settings, and is growing rapidly. AMICA™ is the data foundation that powers our immune models and our therapeutics discovery and development efforts.
Immunai data foundation
Deep, multiomic
Single cell profiling of clinical cohorts
Dynamic
Functional genomics platform for multiplexed perturbations
Broad, curated
Data from public sources
Unified
Clinical and immunological annotations for integrative analysis
5-6k
Studies
350+ Sc studies
800+
Cell types
80 Immune cell types
500+
Diseases
100K+
Patient samples
Curated public transcriptomics data
Immunai has created the world’s largest compendium of deeply curated transcriptomic data, which covers a wide range of research and therapeutic areas. The use of controlled vocabularies, stringent quality control, accurate cell-type annotation, data normalization, and peer-review by domain experts, enables machine learning and interpretation of biological phenomena across organisms and diseases.
Single cell multiomic immune profiling
Immunai single cell profiling platform generates high quality and high scale data assets from biology, capturing key -omics at a single cell granularity that provides a comprehensive view into cell behavior and state. Our ML-driven pipeline translates this massive, high complexity data into standardized cell models that enable analysis across time, patients, and experiments.
A machine learning based immune profiling and discovery platform
Dehashing, Multiplet Removal, and Refined Cell Annotation
Multi-task cell identity neural network and expert system to eliminate contamination and improve cell annotation
Single-Cell Anomaly Filtering
Nearest neighbor graph analysis identifies rare populations
Robust DEG Computation
Single-cell bootstrap and quantile smoothing improves sensitivity and specificity of DEG
Perturbation Prediction
Composable single-cell autoencoder infers gene regulatory networks at scale
Experiment Planning
Multi-armed bandit and optimistic phenotypic regression network
Immunai leverages transfer learning between our ever growing experimental and patient cohort datasets
By applying transfer learning to the immune system, Immunai is able to measure the commonalities that exist between different cell types and disease indications in order to uncover novel insights and accelerate drug discovery and development.
Single-cell analysis supported by big data and machine learning:
- – Cell annotations
- – Differential expression analysis
- – Multi-cohort meta-analysis
- – GRN inference
Functional genomics at single cell resolution
Immunai leverages ML-guided pooled CRISPR perturbations in primary cells to interrogate disease mechanisms and tease out causation at single cell resolution.
Immunai’s platform perturbs hundreds of genes across millions of cells per experiment in vitro with rich single cell multiomic readout.
We create libraries of causative perturbational signatures across a broad experimental landscape and a variety of immune model systems. When mapped against multiomic profiling of our longitudinal patient cohorts, we can connect molecular perturbations to clinical context, and vice versa, to generate novel, differentiated insights for discovery and development.
Predicting perturbations
to guide experimentsSystematic
experimental perturbation to dissect immune cell gene regulationWe design optimized in vitro model systems to better recapitulate human physiology and evaluate drugs
Knowing if and how patients will respond to a new immunotherapy remains opaque.
Many current in vitro systems used for preclinical testing do not adequately model human physiology, which limits their predictive power to accurately evaluate drugs.
Our deeper understanding of the immune cascade allows us to quickly optimize in vitro models. Alongside our data-driven machine learning pipeline, we leverage patient-derived multi-omic signatures to identify those model systems with the greatest predictive power for therapeutic priorities. These precise patient-derived signatures can be translated into predictive biomarkers which inform clinical strategy decisions such as patient selection or indication selection.